Chuck has just written a blog which was very similar to a blog that I was working and I agree with a lot of what he says but I’d take it a lot further and there are some interesting conclusions and potentials along the way which could open the market for interesting innovation going forward.

Chuck talks about storage as being software, go read his blog; there’s little I would disagree with there at all; well, until he starts talking about EMC products! However, I would go much further and suggest that we are getting to the stage where all infrastructure at a very real level is becoming software. Although I am not totally enamoured with the Intel-focussed monoculture, it has allowed a common hardware platform and it is flattening the playing field when it comes to hardware differentiation.

In fact, to differentiate your hardware is going to become increasingly expensive and hard, so why bother? Yes, there will be edge cases where hardware differentiation will be a key USP but in 90% of all use-cases; an Intel box assembled from bog-standard off-the-shelf parts will be good enough.

If we then factor in pervasive virtualisation in the data-centre; we have a platform which has become pretty much standardised and commoditised. I would like to see more ‘standardisation’ in the virtualisation arena but it’s not that bad at present and you really do not have that many choices.

So this has some interesting results; it lowers the cost of entry for new players in the market, if they no longer have to spend time developing a hardware platform and packaging their infrastructure product as a device but simply as a ‘soft appliance’; they can get it out there a lot quicker. If they can now rely on the virtualisation layer providing them with a common way of accessing hardware services; they can develop a lot quicker and they can try things much faster. The cost of failure is a lot less; it also allows integration with other types of infrastructure to be tested without a lot of expense; this is a plus for both the developer but also for the interested Infrastructure specialist.

The speed to market is greatly enhanced and they can get it front of idiots like me who will download the appliance and have a play and it’s not just storage appliances. There are products like Traffic Squeezer which do WAN Network Traffic Acceleration; there are more open-source router projects than you can shake a stick at.

I am slowly building a virtual Data Centre out of open source or at least free products; I want to see how far it would be possible to get. But I’m not suggesting that anyone would do this for real today, although I can see some Cloud providers having a really good bash at it. This approach probably would not make sense for most companies as a complete strategy but as part of a strategy, it may well be worth considering. There are large companies out there who invest in start-ups simply to develop stuff for them to use internally; the advent as ‘infrastructure as software’ without a huge investment in tooling up to build hardware means this a very viable approach for the biggest companies.

Google do it, Amazon do it but often you hear the comments, ‘Well they employ very clever people, so it’s easier for them!’ Well, don’t you employ clever people? Or are you saying that all your employees are second rate.

It took me a long time to come round to the idea that commoditisation and standardisation could drive innovation at all levels; I now believe that it could. It’s not just about more and more Web applications; it could drive a new wave of infrastructure innovation as well.

This leaves some interesting conflicts on the horizon for companies like VMware; perhaps VMware might want to get into infrastructure appliances but that would lead them into direct competition with the Mothership. Infrastructure as software; interesting times.

Here’s some links of things worth looking at or playing with, it possibly includes things which are not strictly infrastructure but are interesting anyway. Some are great, some not so great; some show great potential.

• Traffic Squeezer
• Openfiler
• Samba Ctdb
• Vyatta
• Amanda
• OpenDedup

And of course, there is Ubuntu Server; which will let you build your own cloud for free; there are various ZFS-based storage appliances. You can build your own appliances as well, packaging up and integrating components in the way you want.

One area where EMC have shown great foresight and that is investing in the vSpecialist team and building a team out of diverse specialities because as infrastructure becomes software; the cross-over between the infrastructure disciplines will become even more mandated. Now the vSpecialist team may be very focused on the ‘EMC product set’ but if I was an SI or another vendor playing in this space; I would be looking at doing something very similar in the near future.

© Martin for Gestalt IT, 2010. | Infrastructure is Software
Read more posts categorized as Cloud Computing, Networking, Storage

Having worked in Corporate Infrastructure for many years, fighting the good fight and trying to get the enemy to conform to best practise and generally think beyond the next line of code that they are writing, I surrender! I throw the towel in!

I’d like to take an example without being too specific from my own experience; currently in the race to innovate, many corners are cut. Every screen whether it is a TV screen or the screen on a hand-held games device is now potential target for a product.

Our developers build proof-of-concept systems on servers under their desks at best and often on desktops which might be laying around spare; proof-of-concepts are demonstrated and then proof-of-concepts are products. Cycle times for delivery of new product are often measured in weeks, not the months and years of the past.

No thought is made to non-functional requirements, the capability is all and it’s now a product with paying customers who expect it to be up and running. In fact, the mention of non-functional requirements are enough to send customers running for the hills, they don’t want to know.

But of course, in reality we have to retro-engineer in all those requirements that are required to make it a supportable service. And whilst we are doing this, our people are in a constant struggle to keep this system up. At times, our people do remarkable things in this battle and they should be commended for doing so.

But there must be a better way than PCs under desks and developers acting like MacGyver to get a new service together? I think there is, I think it’s a dynamic, scalable, on-demand infrastructure; call it Cloud, call it what you will. Users should be given the ability to throw-up an environment quickly and easily with almost no thought made to the availability, scalability requirements which will surely be required if the development is a success.

Yes, some use is made of public Cloud but I think that pales into insignificance when compared to use of commodity hardware which is just lying about; this has been going on since the PC reared it’s ugly head but as PC’s have become more powerful with the ability to run server operating systems and especially with the rise of Open Source, every developer has the ability to build a ‘production environment’ without permission.

Are they doing things in Cloud? Yes, they surely are but I suspect it is a more common situation for a infrastructure team to be presented with a skunk-works system built out of commodity and be told to make it live…..tomorrow.

And it is then that the Cloud comes into play, it would be much better to have built an environment which allows developers the flexibility and agility that they require, having them work in an environment which can be promoted to a production environment rapidly. An environment which is as cheap and as flexible as the development teams believe that their PCs are but gives the infrastructure teams the supportability that they require.

Non-functional requirements whether we like it or not are just bolt-ons; after-thoughts. They are seen as the obstacles which stop customers gettting the new services which they require, it’s time to move on. I still believe that developers should pay attention to non-functional requirements, I still believe that systems should be designed with availability, scalability, recoverability etc in mind but I think that this is now needs to provided in such a way that it is transparent to all.

Of course this transparency should be completely transparent and open allowing rapid migration between providers, private and public…but that is another story!

© Martin for Gestalt IT, 2010. | No Longer Functional
Read more posts categorized as Cloud Computing

Okay, so NetApp have announced the NetApp StorageGrid product, however at the moment as far as I can see it is a simple rebrand of the ByCast product. I am not sure whether I was expecting anything more or whether I was expecting them to go dark with the Bycast product set for the time being whilst they work out what the hell they are going to do with it and at least come up with an integration strategy for the products.

Like many I wonder what this does to the whole Unified Storage message because NetApp now have two disparate storage product sets which are not integrated; I’m sure that they are briefing the integration message under NDA and if not, I’d ask why? But I’d interested to see what form the integration takes, will be it be at the tools level or will be it more fundamental integration more akin to OnTap 8.

As NetApp have announced it under the Storage Management Software product set, it appears to be the former, certainly for the short to medium term and I suspect that NetApp are going to be very wary about going after a full blown integration or at least a public statement on it after the torturous integration of Spinnaker.

The data-sheet shows a software gateway layer sitting above the OnTap filers, well I think that’s what it shows. It says that the front-end app server supports NFS/CIFS/HTTP(Restful) protocols communicating with the back-end storage via NFSv3; so theoretically, the back-end storage could be anything supporting NFSv3? But at present the data sheet actually shows a very restricted storage environment supported, namely FAS31x0 and FAS20xx and only SATA drives, so there seems to be no way of utilising your legacy storage in your StorageGrid. This is a little disappointing but no huge surprise, if EMC decide to ‘support’ third party storage with Atmos, it should be no biggie for NetApp to follow suit with StorageGrid; or perhaps vice-versa.

And as ByCast StorageGrid was resold by a number of other vendors, what is the ongoing roadmap for those customers who are running StorageGrid with different vendors storage behind it? Are these customer’s going to be expected to move to NetApp storage?

Also from the diagram in the data-sheet;

‘NetApp StorageGRID object-based storage solution brings the best of NAS and RESTful HTTP client access together’

Now I am willing accept that NetApp claim that the Filer product set are the best of NAS but to provide this ‘best of breed functionality’ with the StorageGrid product would imply a deeper level of integration than I can currently see or are they claiming that the Bycast product was actually the best NAS product out there?

Is the Filer behind the Gateway being treated as pretty a dumb-share-only Filer and not leveraging any of the OnTap features at all? Even if this is the case, it is a cute move politically as the sales-team will not see any potential Filer sales being cannibalised by this new product. A problem that I believe that EMC might have had to deal with the Atmos product set.

One of the keys will be how NetApp present the integration; will they add StorageGrid to Ops Manager? It seems to make sense that you add it at that level because Ops Manager is the preferred way of managing multiple Filers and to get the most out of StorageGrid, there will be many Filers.It also keeps it in the realms of the familiar.

If it is seen as very much a different product it makes the Unified Storage pitch a little harder as it becomes mostly-Unified-Storage product which is a bit like being a little bit unique.

So this announcement asks many more questions than it answers!

And one final comment, what is the difference between an Storage Grid and a Storage Cloud? Is it an Object Cloud or an Object Grid? Does the Object Cloud live in the Storage Grid??

© Martin for Gestalt IT, 2010. | NetApp StorageGrid – More Questions than Answers?
Read more posts categorized as Cloud Computing, Storage

This week’s Tech Field Day roundtable discussion is an audio chat on the subject of stack wars – the creation of all-in-one “stacks” of IT equipment by the major vendors. These stacks have been quite a hot topic of discussion both inside Gestalt IT and in the community as a whole.

The discussion was wide-ranging, and we’ve split it into two episodes. This second episode includes discussion of the following:

• HP’s unique position as a soup-to-nuts infrastructure supplier
• The enterprise readiness of 3COM, D-Link, Virtual Connect, Xsigo
• Cisco’s secret plan with UCS marketing
• The concept of integrating the VMware hypervisor as firmware in the stack
• The exact definition of stack – what does it really mean?
• Is there a difference between a “marketing stack” or reference architecture and a “real stack”?
• What is the compelling economic driver that makes people move to vertical integration?
• What is VMware’s long-term strategy for stack wars?
• Was vBlock designed for the cloud?

Joining the call are the following Gestalt IT authors and Field Day delegates:

• Jason Boche (@JasonBoche, Boche.net)
• Rich Brambley (@RBrambley, VM /ETC)
• Carlo Costanzo (@CCostan, VMware Info)
• Chris Evans (@ChrisMEvans, The Storage Architect)
• Greg Ferro (@EtherealMind, EtherealMind)
• Edward Haletky (@Texiwill, The Virtualization Practice)
• Joerg Hallbauer (, Joerg’s Storage Blog)
• Robin Harris (@StorageMojo, Storage Mojo)
• Rod Haywood (@Rodos, Musings of Rodos)
• John Hickson (@StudioSystems, Studio Sysadmins)
• Scott Lowe (@Scott_Lowe, ScottLowe.org)
• John Obeto (@JohnObeto, Absolutely Windows)
• Devang Panchigar (@StorageNerve, StorageNerve)
• Bas Raayman (@BasRaayman, Bas Raayman)
• Ed Saipetch (@EdSai, Breathing Data)

Download the stack wars roundtable podcast now, or subscribe in iTunes to tune in to the discussion!

© Stephen Foskett for Gestalt IT, 2010. | Podcast 7: Stack Wars Roundtable 2
Read more posts categorized as Cloud Computing, Exclusive, Featured, Server Virtualization, Tech Field Day Roundtable Podcast, Top Story

Before Google could even take to the stage to announce their new “Google Storage for Developers” cloud storage offering in their I/O conference keynote, Amazon hit back with a new low-cost “Reduced Redundancy Storage” option for S3. The titans are at war, and cloud storage is the new battle ground. But what was really announced? And should you care?

Defining Durability

Amazon begins their RRS rollout with an apt discussion of durability. To quote:

Let’s define durability (with respect to an object stored in S3) as the probability that the object will remain intact and accessible after a period of one year. 100% durability would mean that there’s no possible way for the object to be lost, 90% durability would mean that there’s a 1-in-10 chance, and so forth.

I like this turn of phrase, and encourage other storage vendors to consider it as well. It’s especially appropriate for a discussion of public cloud storage, since the traditional SLA (“availability”) relies on many systems between data and application. There are many failure scenarios that can result in a loss of availability without any risk of data loss, and Amazon is wise to make the distinction here.

Also see Amazon CTO Werner Vogels’ discussion of S3 durability

Amazon claims that “the durability of an object stored in Amazon S3 is 99.999999999%. If you store 10,000 objects with us, on average we may lose one of them every 10 million years or so. This storage is designed in such a way that we can sustain the concurrent loss of data in two separate storage facilities.” That’s eleven nines – a Very Large Number. RRS changes this: “Objects stored using RRS have a durability of 99.99%, or four 9’s. If you store 10,000 objects with us, on average we may lose one of them every year. RRS is designed to sustain the loss of data in a single facility.” Eleven nines remains the default for data written to S3, but an API call can set the REDUCED_REDUNDANCY flag on demand.

The idea is that some data can be regenerated or simply requires less “durability”. The value for the customer is reduced cost: RRS is 1/3 cheaper than regular S3 storage at $0.10 per GB per month to start, a substantial discount designed to counter Google’s forthcoming offering.

Stephen’s Stance

Amazon S3 RRS is a risky move, but the company handled the announcement exactly right. Rather than focusing on the reduction in cost, they turned the spotlight to the built-in reliability of their existing offering. Yet it’s a fact that not all data is equal, and much could be stored with less durability and at a lower price, and Amazon’s new option enables this. The company also made the right move by making reduced-reliability an opt-in option. RRS is a good move for Amazon, and I expect competitors (Google, Rackspace, etc) to follow.

© Stephen Foskett for Gestalt IT, 2010. | Amazon Redefines Durability with Reduced Reliability Storage (RRS)
Read more posts categorized as Cloud Computing, Storage

So a while back, Cisco, VMWare and EMC announced that they are forming a partnership to co-operatively sell and support products. This joint venture is named Acadia and is made up of a specifically screened and selected sales people and engineers from each company. What is particularly notable is that EMC and Cisco have committed in excess of $200 million to make it work. This tells me that they are serious about this project.

Then they added some of their best sales people to the mix including engineers and then went knocking on the doors of every large company in the world in a “no large customer left untouched” blitzkrieg of one on one marketing.

So What is Acadia made of ?

Acadia is offering a package of actual products (not PowerPoint releases) from EMC, Cisco and VMware using a marketing strategy called Vblock. At Boston Gestalt IT Field Day, we received an excellent presentation on the details around the Vblock package from Ed Sai and Scott Lowe that sparked a heated response. I felt that the key issues were around two points:

• Why are the Vblocks so Large ?
• What happens if you exceed the operational parameters of a Vblock ?

Because there was much discussion about why the Vblock that have some strict limitations, I wanted to look into why these limitations exist and why people would object so vehemently to them. And also talk around the question for why Vblocks are big. Too big for most people to buy.

Comprehending the Limits on the joint venture

Acadia is not a merger (yet) of EMC and Cisco, but a partnership focused on the Data Centre. The partnership only promotes selected product groups from each company and even specific products within those group. For example, a Vblock does not include a Nexus 7000 switch, only the Nexus 5000 / 6000 / 2000 as the edge switches. ( You would use your own backbone to connect to your Ethernet network and legacy FibreChannel network ). VMware does not offer VDI solutions. EMC offers CLARiiON in Vblock1, and Symmetrix in Vblock2.

These limits may mean that your choices are limited. Some people regard this as as major concern. I regard this as a reasonable trade off – good service, easy marketing against lack of choice and flexibility.

Other Companies

It’s worth remembering that the joint venture is not exclusive. Both EMC and Cisco are free to pursue partnerships with other companies, and both are actively seen partnering in the market with other companies. Thus recent Cisco / NetApp and EMC/IBM announcements could create a confusing picture for some people who don’t understand these dynamics. That said, from what I’ve seen the Acadia partnership seems to be a “big deal” for both sides with a lot of passion, commitment and, most importantly, significant amounts of cash that fund expensive marketing campaigns. In this case, marketing that seems to actually be relevant and useful.

Selected Products

Acadia only works with selected products from both Cisco, EMC and VMWare. That is, they only offers products from the Data Centre Business Unit from Cisco, selected products Centerra and Celera at EMC and VMWare vSphere. The product is positioned in the market as a private cloud so any technology not relevant to that message isn’t going to be considered eg. No firewalls, no edge routers, no wireless, no backup software, no deduplication,.

My perception is that clear business goals drive this logic:

• Market clarity – limiting the product set allows Acadia to have a highly targeted product message, especially with the “cloud” buzzword attached.
• Technical capacity to deliver an effective sales force
• Leaves space for ‘other partners’ to have similar relationships with customers and not ‘upset’ the reseller marketplace
• minimise the business impact if the project fails
• customer want guarantees that it will work which should get better sales results.
• Acadia has to deliver the solution and deliver the support promise. Not easy when so many technologies are involved, with so many competing interests

I particularly want to focus on the Technical Capacity and Customer Guarantees elements of the argument. If you figure that EMC, Cisco and VMware are bringing a group of vendor engineers together, think about the problems you might have: engineers may have little or no expertise outside of their own company or worked in a cross-functional team. To overcome these problems I would go for a narrow product selection and then expand it over time.

Remember that large companies create significant opportunities for ineptitude and incapacity. Layers of management can create indirection and misalignment just as easily as they can create opportunity and focus. Bringing together teams from three large companies could easily go wrong (case study – HP and it’s many divisions that are rarely coherent).

There are other topics, such as upsetting the resellers, market clarity are all pretty obvious and shouldn’t need discussion.

What are Vblocks ?

The Vblock components are the cumulative permissions of what the marketing and business limitations will allow. There are no technical limits to what equipment could be in a Vblock except for the ability of Acadia to provide a good level of support. These limits are somewhat onerous but not entirely unreasonable. There are currently two Vblocks defined (and more are planned) – known as Vblock 1 and Vblock Two, they are designed, tested and (as is the fashion) certified to within certain, rigidly defined performance levels.

Basic Components

The Vblock architecture is reduced to five layers of physical infrastructure and roughly shown in this diagram taken from “Vblock Infrastructure Packages Reference Architecture” released in 2010.


Outline Bill of Materials

From this, you can see there is a fair amount of variation within the Vblock parts list to cater for your specific requirements. To quote Acadia :

A Vblock consists of a minimum and maximum amount of components that offer balanced I/O, bandwidth, and storage capacity relative to the compute and storage arrays offered. Each Vblock is a fully redundant autonomous system that has 1+1 or N+1 redundancy enabled by default.

In practice, Vblock 1 storage is scaled to deliver from ~42 TB/41000 IOPS (Min) to ~68 TB/ 50000 IOPS (Max). Vblock 2 storage is scaled to deliver from ~140 TB/92000 IOPS (Min) to ~211 TB/140000 IOPS (Max). The hardware is selected to match your performance requirements.

Other flexibilities

Some other quotes from the Cisco / EMC / VMware reference architecture guide:

For Vblock 1, there are no hard disks on the B-200 series blades as all boot services and storage are provided by the SAN. Hawever a snall hard drive may be installed if local page memory is required for vSphere.

For Vblock 2, each B-200 series blade module has 72GB SATA drive for page memory purposes. If required, these can be removed to reduce power and cooling overhead, increase MTBF or save costs.

The MDS9506 switches are recommended, but may optionally be changed for 9509 or 9513 to scale capacity or reduced to an MDS 9222i is less density is required

Inflexibilties

For Vblock 1, there are no hard disks on the B-200 series blades as all boot services and storage are provided by the SAN. Hawever a snall hard drive may be installed if local page memory is required for vSphere. If the local disk is user for main storage or operating system storage, it is not considered a Vblock and is a custom implementation at this point.

It is also acceptable for operating system and applications to be run directly on the B-200 series blades. It should be noted that other hypervisors are not supported by Vblocks and invalidate the Vblock support agreement (my emphasis)

The Limits are arbitrary, practical and mainly marketing

At the Boston Tech Field Day, we saw a number of questions on why there are limits, and what happens if you exceed the solution parameters. I believe it is a simple marketing reason – simple products, simple factors, easy to comprehend makes for a product that is easy for sales to sell, and even easier for customers to buy. Pricing is clear and unambiguous, and sales peoples will attempt to focus the customers away from price onto other topics. Like low power, special UCS features, ease of operation, etc etc.

The key factor is that the Cisco UCS server will be perceived by many as unproven. You can be sure that HP / IBM / Dell are going to work that Fear / Uncertainty / Doubt in a competitive situation.

Exceeding the Limits – You Certainly SHOULD

If you choose to exceed the offered the solution, then that is a simple choice. It’s not something to be particularly concerned about. Currently, we all build stacks of servers, storage and networks and provide our own support. Why would buying a guaranteed stack make any difference ?

These limits are to ensure that service guarantees can be met. The service guarantee exists to help customer to buy these products., it does not exist to make the products work. However, when the CIO is about to approve a purchase order for half million pounds they will be looking for a comfort level that it’s going to work. Those Cisco UCS servers are new and you want some assurance.

So once the system is working, feel free to head out in your own direction. Meeting the requirements of your business is more important that having a an end-to-end guarantee in overall picture. Remember, Cisco / EMC / VMware is still going to support each product using the same process that we have today.

The Fear of Lock In

My concern with the Vblock bundle is fear of lock in. The Cisco UCS product is a closed solution within the racks. Across the backbone of the Data Centre will also be Cisco Nexus 7000 switches for “best compatibility”. All server components must be purchased from Cisco who will be able to arbitrarily set the price. All and any server upgrades, and extra server must come from Cisco.

All storage vendors are an effective lock in today. The so-called “certification programs” ensure that EMC and other “chosen few” are restricted to the supply of drives and accessories.

If Cisco / EMC achieves a dominant position then they have the ability to control pricing. Since both companies are already dominant in their respective markets, and tend to “reassuringly expensive” price models, this is a well founded concern.

What does a Vblock physically look like ?

While every Data Centre is different, every rack looks pretty much the same. During my research, I have realised that a Vblock takes very little space in the data centre. In fact, it’s not until you see the amount of space that you realise the impact to

Vblock 1 – Racked

This is a front view of the Vblock 1 with storage, servers and switches.

Caption Text.(Click for a full size image)

And this is the rear view. Notice how few cables are used for physical connections here.

Caption Text.(Click for a full size image)

Vblock 2 Racked

A Vblock 2 is quite a bit larger and needs more storage. The server blades are larger (more CPU, more RAM per blade). I believe that the increased IOPS for the storage subsystem means that bigger storage units are needed. I think five racks is at the smaller side of the Vblock 2 possibilities.

Caption Text.(Click for a full size image)



Realising the dream – no cables

Caption Text.(Click for a full size image)

Vblock expansion

Even though the Vblock is modular, you can bet that Acadia has a nice little pitch around expansion. Sure enough, it is spelled out in the Architecture Reference:

Caption Text.(Click for a full size image)

As Vblocks are adde, the capacity of the Vblock scales either as an aggregated pool, whereby any ICS blade can access any storage disks n the SAN or as an isolated silo. For example, it is perfectly acceptable to aggragte to Vblock 1′s to provide capacity for 6000 VMs that can share common storage capacity… As long as storage capacity is added in conjunction with compute capacity to maintain balanced performance as published within the Vblock, the system does not require any additional validation .

Heh, you bet they want to sell you lots more so of course there is an answer for this. Once you have made the first purchase, you are likely to come back for more. Lets make it easy for the customer to do that.

So, Why are vBlocks so Large ?

A Vblock 1 is expected to support from 800 to 3000 VMs, and a VBlock 2 is expected to support 3000 to 6000 virtual machines. Since I started writing this post, a Vblock 0 has been announced that supports 300 to 800 VMs.

You might think that 800 is a lot of VMs. You might think that the price of a Vblock is too much. In which case, you aren’t someone who is going to buy it. If you can’t scratch together the several hundred thousand in single purchase order, you do’t understand the question. Today, many companies buy infrastructure as part of the project. If a project needs a server and a switch then that is funded. If a project needs a data centre, then a data centre is built. If a project needs fifty servers, then they MIGHT consider using VMware. Mostly we wait a few years and then have a big project to clear up the mess created by project funding of infrastructure.

Cisco and EMC don’t want to sell this product to small business. Cisco in particular wants to establish themselves in the marketplace as quickly as possible. Therefore selling to big companies, who are doing the “big cleanup project”, who are taking the time to evaluate the whole cost of ownership cycle, implementing with a full cycle approach is what Cisco wants. This is important.

Why ? Because Cisco wants to be able to produce white papers claiming the “huge cost reductions”, and “radical operations alignment” and “enabling technologies” etc etc. These white papers have real impact on the analysts and customers who figure that if they buy the product they will get those benefits. All these great stories, so loved by ‘journalists’ convince the middle sized market to buy Cisco servers.

What does EMC get ? EMC gets to move into the mid-sized market right alongside Cisco. EMC has been targeting the high end of the market, and they need new markets. A move down into the mid-sized market must not damage the high end products and branding (since this could affect profits), and attaching to a “differentiated” product will keep the premium marketing message intact.

EtherealMind View

Basically, I view the Acadia partnership as an upmarket fancy marketing campaign with extensive tie-ins. It’s “dressed up with a little black dress and killer set of heels” and well funded to boot. This type of thing normally makes me distrustful as many of these initiatives have turned out to be one night stands with an uncomfortable and unhappy morning after.

On the other hand, both Cisco and EMC have attached their corporate goodwill to this program in a major way. If Acadia fails, both companies will suffer significant image loss. It’s not marriage, but they are definitely living together.

The Challenge to the Status Quo

One of the most exciting (to my mind) aspects of the Vblock concept, is the potential to change the purchasing cycle of IT Projects. Many companies fund new purchases from projects only. This tends to create an inherently failed infrastructure buildout where each project buys what it needs. Project “cost controls” mean that infrastructure is rarely purchased to outside the project requirements. That is, buying additional disk drives or edge switches is easy, but buying additional arrays or core switches is almost impossible.

A Vblock could represent a “once a year” purchase of server infrastructure. Much easier than endlessly integrating, and reintegrating a hodge podge of servers, memory, disk drives and accessories into some sort of complete system. Compare this with the HP C-Class chassis concept which requires endless upgrades to scale into new and more useful configurations.

Most importantly, once engineers are freed from mundane tasks of installing memory and disk drives, they can move to focussing on the management platforms and dramatically enhancing the business value of computing. Tasks such as automating software and OS deployment, automated failure responses, performance analysis and

Cisco needs to buy EMC

Ultimately I remain convinced that Cisco must and will buy EMC to compete with HP and IBM. The Acadia joint venture looks a lot like a trial or temporary position while the market reaction is gauged. EMC has a marcap of USD$34 billion and revenue of USD$14 billion while Cisco has marcap of USD$154 billion on revenue of USD$35.5 billing. Cisco appears capable of acquiring EMC and is believed to have sufficient cash reserves to make a substantial cash rich offer. EMC probably has pretentions to be a lot bigger and don’t want to be bought right now. EMC egos are well known for their ‘capacity’ and they work hard at that. Combine that with a Wall Street that is anti-acquisition at the moment, and Cisco can’t make a move right now.

Public Clouds are not ready for the Enterprise

Public Clouds are fine for people that are not conducting serious business today. Dot Com startups, marketing pushes, publicity campaigns and other low vitality requirements could be considered for cloud infrastructure due to their trivial nature, but serious business applications are unlikely to move into the cloud with the current level of capability.

However, private clouds are practical today. To me, Vblock looks like an early mover in the Private Cloud space for IaaS. The option for a corporate IT department to build a three rack system that hosts between 300 to 700 servers is an exciting development. If I can reduce the amount of time my operations teams spend doing trivial and repetitive tasks and move them into systems automation on the management platforms, then we are moving to better place.

Technology Acceptance

There are no technology issues in this stack. In general, people perceive that Cisco and EMC are market leaders in their respective spaces, and the corporately the two companies fit well together. Unlike other stacks, there doesn’t appear to be a product that is less than excellent. In this sense, few people will dismiss the Acadia solution on technology grounds.

Conclusion

Convincing IT Management to take on the change put forward by Cisco / EMC is going to require a lot of marketing and sales. And that’s exactly what Acadia is all about. The technology takes care of itself here, Acadia is just marketing exercise, but a good one.

© Etherealmind for Gestalt IT, 2010. | Market positioning Acadia, EMC, Cisco and the whole vBlock Idea
Read more posts categorized as Cloud Computing, Featured, Networking, Server Virtualization, Storage, Top Story

EMC is the 800 lb gorilla of the enterprise storage industry, but the company has much bigger plans. Although CEO Joe Tucci kicked off his keynote by claiming “We are an infrastructure company and proud of it,” EMC’s ambitions must go way beyond IT infrastructure. Each acquisition and strategy announcement is an attempt by EMC adapt to a fundamentally-transformed enterprise IT world.

The Non-Box Company

EMC moved out of the “box company” role in 2003, purchasing Legato Systems to move into enterprise backup software and Documentum for enterprise content management, but this was only a warm-up. Many were puzzled by the 2003 purchase of virtualization pioneer, VMware, but it has played out to be one of the shrewdest moves of the decade. Although their revenue remains small in comparison to the parent company, VMware is arguably more important today to the modern data center than EMC itself. Later purchases like System Management Arts (SMARTS), Configuresoft, and RSA further expanded EMC’s footprint.

More telling are the moves VMware has made. SpringSource, Zimbra, and GemStone add up to very much a non-infrastructure world for the “infrastructure company.” EMC is clearly moving into the application platform space, leveraging this reach to consolidate their control over the underlying infrastructure.

Moving On Up

This seems to be the real EMC vision: Move up the stack, offer a compelling IT strategy, and move tons of heavy datacenter gear. This isn’t unique; It’s a re-play of IBM’s game plan from decades past. But EMC finds itself in a moment of transformation for corporate IT: Application development teams no longer care about infrastructure, and the rest of IT must finally transform itself to keep up with their needs.

Read my thoughts on The Techie/Business Schism for more on this IT transformation

Make no mistake: IT is in the middle of a revolution, not a simple equipment transition (think mainframe-mini-micro). Rather, the new world of IT is application-centric and big enterprise iron is becoming less visible and strategic. Surely, future IT architects will be able to deploy monolithic integrated blocks or stacks (like the EMC/Cisco vBlock), but many will choose to deploy lighter scale-out systems based on commodity hardware. The choice will be made based on finance or business strategy, not technological capabilities because higher-layer software will make hardware increasingly irrelevant.

Selling the Strategy

What is an avowed infrastructure company like EMC to do? Many of their competitors are doubling down on advanced storage hardware, but the smart ones are focusing on software instead. EMC is certainly moving towards a common hardware platform differentiated by software “personality” but their strategy must go well beyond this. The company must focus on the new IT world, selling product as a foundation for the next generation of applications, not merely a “faster/better” storage box for today’s apps.

Yet it would be foolish to make a wholesale move and abandon existing systems and protocols. EMC’s revenue stream depends on continued Fibre Channel, iSCSI, and NAS array sales even as they transition to systems that bear little resemblance to today’s Symmetrix or Celerra. They must develop and market a forward-looking strategy yet not abandon their cash cows.

In other words, EMC must say one thing while, for the most part, doing something entirely different. This “run the business while changing the business” challenge has been the downfall of many. The few who have successfully transitioned are the exception rather than the rule. EMC has a tough road ahead, but my conversations with company insiders at EMC World this week show that they are well aware of the challenge. This puts them well ahead of many in the storage industry, and explains my continuing focus on EMC.

© Stephen Foskett for Gestalt IT, 2010. | EMC’s Tough Road to the Post-Infrastructure Future
Read more posts categorized as Cloud Computing, Storage

Why is this happening?

I think a number of factors have lead to the “stack” or the return to the mainframe model.

One of these is virtualisation. Virtualisation has in many ways collapsed the different components of compute, networking and storage into a blob where differences in each individual layer are diminished. The layers are abstracted away through the hardware independence and it starts to make sense to obtain an entire “virtualisation” machine rather than build it from scratch each time.

Likewise the continue performance improvements means that these more generic solutions can solve a might greater scope of workload than ever before. Certainly the advancement in hardware processing capacity has outpaced softwares ability to consume it.

The second element is the change in behaviour of the vendors. The vendors are much more willing to get in front and sell to end users these days. Even those vendors who are so channel focused do this. I remember 10 years ago in the integration space; it was only the top of town that could get a sales person or a system engineer from a vendor to visit and flaunt their wares. These days, if you are a fish and chip shop you could probably get a vendor to turn up for a presentation and a proof of concept.

However the vendors are realising what the system integrators learnt a long time ago, there is money to be made by combining all the parts and providing a whole solution. Plus the cost of sale can drop if you can have a few packages that fit most sales, rather than doing everything as customised solutions.

The vendors have realised that by using virtualisation and stacks they can make larger sales whilst reducing their cost of sales whilst targeting an increase number of opportunities.

Is this good for end-users?

It is probably a little early to see how the benefits for end users will pan out. If we learn from history the mainframe era certainly had its drawbacks for many organisations towards the end.

There are certainly some up sides for the end users. Improved levels of support and integration can’t be bad. Certainly lower costs by removing lots of customised design work and through economies of scale is going to be a benefit.

But there will be drawbacks too. I don’t think anyone has been thinking or talking about the lifecycle of these stacks. Will you have to replace the entire stack each time at end of life to keep your support? What if you are happy with your compute and storage but have brought in a new networking fabric that is much faster, do you have to throw the baby out with the bathwater?

Where are IBM and Dell?

My hunch is that IBM and especially Dell are in the wings waiting to see how things pan out. Let the early adopters play and make the market, see where the success and failures are. Once they have learned from all of their mistakes they can swing in bringing in their existing value propositions. They don’t want to leave it very long but we are still in the incubation period for the stacks.

I think Dell is the one to watch here rather than IBM. After all they have been essentially doing this today in the lower end of town. They have the parts to bring it together and they can suck the bottom tier right out of the market from under everyone else.

Of course HP is the other one to watch very carefully, especially now that 3Com is on board.

What about the smaller players?

If the stacks take off the smaller players are going to continue to do what they have always done, value add. Many elements of technology have turned to commodity. Remember the days when you would always pay people to come in and do Exchange deployments. Remember how hard it was to maintain Unified Communications solutions. Over time the technologies became utility enough that organisations could do these themselves.

Smaller players, whether they be systems integrators or vendors will continue to find those niche requirements and the hard projects or problems that will always be around. Our consumption of technology is growing not shrinking, the small players will continue to deliver and support the early adoption technologies.

What does this do to innovation?

Many people in IT need to understand that we are in a young industry which is starting to mature. A lot of stuff really is going to become utility and standardised, thats the way industries evolve. Yet innovation still continues in mature industries.

What about cloud?

Bingo, we can’t talk about anything with mentioning Cloud. Are the stacks the vendors means of abating the move of all of the revenue to a handful of global cloud providers. Build an internal Cloud with our stack please so we can keep making some money off you.

I see the stacks working well with the Cloud, certainly the IaaS and PaaS based ones. Improved standards adoption will allow the federation and creation of meta Clouds. So there is still a place for internal work loads (there is some good thinking on this that uses the commercial property market or food production analogies).

What I DO predict we will see is the vendors offering the stacks to Enterprises is Cloud based service models for internal use. How do people consume photocopiers and printers today, they pay per page. The vendor puts in the equipment, maintains it and supplies it, you just pay for what you use. In the future we are going to see this model start to appear more in IT, once you have a utility stack, this can successfully be achieved for the vendor and the Enterprise. By the way virtualisation is the key enabler here, abstracting away all the hardware from the workloads.

Is this the ultimate form of IT infrastructure?

The word “ultimate” makes it sound like the most amazing. No, I don’t think its going to be the ultimate. I think the stacks might become the boring utilities they they are meant to be. A standardised, reliable, cost effective computing block thats does what it is tasked to do, no more and no less. Its the IT version of the multi-function photocopier. Roll them in and roll them out. The stacks are more about an operational model than speeds, feeds, dials and knobs.

© Rodos for Gestalt IT, 2010. | Rodos on Stack Wars
Read more posts categorized as All, Cloud Computing, Featured, Server Virtualization

The CEO Shuffle is in full force in the IT infrastructure industry. Geoff Barrall, founder and CEO of Data Robotics, recently stepped aside for Tom Buiocchi, and Alex Bakman of VKernel shifted to CTO, bringing in Doug McNary. Now Panasas and ParaScale are replacing their CEOs with industry veterans. The message is clear: The economy is improving, and investors want their portfolio companies to begin growing again.

For more thoughts on the startup growth process, see my related article, “The CEO Shuffle“

Scaling Out

Panasas, of Fremont, CA, is a provider of network-attached storage (NAS) systems and was founded in 1999 by the legendary Garth Gibson, known as the father of RAID. ParaScale, right down the road in Cupertino, sells a cloud storage software layer and was started five years later by Cameron Bahar of Scale8 and Teradata.

Both of these companies are in the scale-out business, enabling massive file storage systems for the largest applications. Panasas is more traditional, relying on NAS protocols and a clustered file system for conventional applications. ParaScale stepped into the cloud storage territory, combining Linux-based storage nodes into a file store with HTTP, NFS, FTP, and WebDAV accessibility.

The two companies have also seen key competitors scooped up in the last year: Panasas shared the scale-out NAS market with IBRIX (now HP), Exanet (now Dell), and ONStor (now LSI). Earlier this month, ParaScale watched Bycast go to NetApp.

Panasas Names Pairman

Panasas has named Faye Pairman, formerly of 3ware and AMCC, to replace Randy Strahan as CEO. Although Strahan held that position for years, the Register suggests he was a bench warmer brought in by investors, Mohr, Davidow Ventures. The Pairman press release supports this, calling Strahan “interim CEO,” though that term was not previously used by the company.

With IBRIX, Exanet, and ONStor gone, the scale-out NAS musical chairs game is winding down. IBM has its own scale-out NAS offering, SONAS, which is based on their GPFS technology. Symantec released FileStore, their own scale-out NAS product, in October of 2009. HP seems happier with IBRIX than they were with PolyServe, and Dell and LSI won’t likely need any more technology. NAS leader, NetApp, has finally integrated the scale-out technology they acquired way back in 2003 from Spinnaker, and EMC is busy integrating their Celerra NAS and Clariion SAN lines. Perhaps the new SGI or Oracle-nee-Sun are shopping?

But perhaps Panasas has an IPO in mind instead. They claim profitability, and that market is beginning to heat up! And the market is kinder now – witness this quarter’s success for public scale-out NAS vendor Isilon (NASDAQ:ISLN).

ParaScale Gets Fehrnstrom

ParaScale has made no official announcement as of yet regarding their CEO change, the company’s web site shows that Sajai Krishnan is no longer CEO. His replacement is Ken Fehrnstrom, who is said to have managed “business development activities inclusive of acquisitions, equity investments, divestiture of non-core business assets, and partnership programs” at Ascend Communications. Interestingly, the ParaScale bio omits that Fehrnstrom is also the former President and COO of managed storage provider StorageNetworks, a prototype for today’s cloud storage service providers.

Beth Winkowski, ParaScale’s PR contact, confirmed the change in leadership in an email message:

ParaScale has appointed a new CEO, Ken Fehrnstrom, for the next stage of its growth. As ParaScale gains market traction, Ken as a multi-time CEO, stands ready to rapidly scale the business.

Sajai cultivated industry-leading recognition for ParaScale in record time, delivered its first product to market and seen it through several significant new releases, and earned the confidence of many brand-name customers and partners.  He has served as CEO for nearly two years.  Sajai has assisted Ken in his transition before leaving the company.

Note: We initially overlooked the mention of Fehrnstrom’s StorageNetworks tenure in the ParaScale bio. It is, in fact, mentioned there.

© Stephen Foskett for Gestalt IT, 2010. | Panasas, Parascale Shuffle CEOs For Growth
Read more posts categorized as Cloud Computing, Exclusive, Featured, Storage, Top Story

Just as public cloud computing is beginning to catch on, the enterprise data center world has been shaken up by the biggest IT product vendors. Rather than sit back and watch their wares commoditized, companies like Cisco, EMC, HP, and now HDS are stepping up to the plate with integrated “stacks” that include server, storage, networking, and management software. The next-layer players, VMware and Microsoft in particular, are joining hands, too, eager to support these stacks.

But what does this all mean to enterprise IT? Who benefits from these vertically-integrated stacks? Who stands to lose out? To paraphrase the wise Jedi master, Yoda, “cloudy, the future is.” So, the stack wars have begun!

What’s a Stack?

The largest IT infrastructure vendors are aligning their own products and joining together in alliances to create vertically-integrated “datacenter in a box” stacks that include server, storage, networking, and higher-level software for virtualization and management. Cisco and EMC are certainly at the forefront of this development with their Acadia VCE vBlock, as demonstrated and deconstructed at our Tech Field Day earlier this month. The reinvigorated HP has its own offering in this space, and is unique in its ability to use all in-house hardware, and IBM and Oracle/Sun are serious challengers. Today, Hitachi announced their own Unified Compute Platform as well.

The battle of the stacks extends upward into software as well. Although VMware and Microsoft (and don’t forget Citrix) are quick to point out that they can run on anyone’s hardware stack, they are being pulled into the war through alliances, ownership, and old feuds. VMware forms the “V” in the Cisco/EMC “VCE” stack, though they also linked up with Cisco and NetApp in February. HP and Microsoft signed an even bigger agreement in January, suggesting that Hyper-V is their preferred virtualization solution. Hitachi’s UCP digs deep into the Microsoft solution set and mixes in home-grown software to do everyone else one better. But perhaps it will be Oracle/Sun who will deliver a real software-focused stack, given Oracle’s historical focus.

All of these stacks have a few things in common:

1. Standard hardware configurations are specified for ease of purchasing and support
2. The hardware stack includes blade servers, integrated I/O technology, Ethernet networking for connectivity, and SAN or NAS storage
3. Unifying software is included to manage the hardware components in one interface
4. A joint services organization is available to help in selection, architecture, and deployment
5. Higher-level software, from the virtualization hypervisor through application platforms, will be included as well

Who’s Driving?

Although all of these stacks include software components, it is interesting to note that it is the hardware guys who are in the driver’s seat. Microsoft and VMware are surely wary of tying up too closely with any one server or storage vendor, lest they lose the ability to compete in other environments. One expects these two to announce partnerships across a few hardware stacks, though it is unlikely that they will both fully integrate with every offering.

When it comes to stack hardware, only HP has a complete solution, having recently acquired 3COM, LeftHand, and Ibrix. IBM, Dell, and Oracle are thin on networking technology, though their storage story is stronger than many give them credit for. EMC clearly lacks the server and networking components and so tied up with a complementary Cisco who is strong in both areas. Then there is Hitachi, with a solid reputation in storage and strong but relatively unknown blade server offering but no clarity on the networking side. Perhaps HDS is already on the phone with Xsigo, VirtenSys, or Arista?

Certainly, all of these hardware vendors would like to offer a complete solution. This leads one to assume that many of the smaller hardware companies will soon be acquired, since they cannot hope to offer their own stack. The networking and virtual I/O companies are especially juicy targets, but don’t count storage out either. One could certainly imagine Cisco picking up a storage company or two and leaving EMC at the altar! Compellent is undervalued right now, and 3PAR must look pretty good, too.

The Cloud Angle

What doe these stacks have to do with that other megatrend in IT, cloud computing? There are two angles here:

1. Integrated stacks are the perfect foundation for private and public cloud deployment, providing Infrastructure as a Service (IaaS) as a turnkey purchase. I called this the “home theater in a box” concept at Tech Field Day: A single line item gets you all the components you need to roll out a large-to-huge virtualization-ready infrastructure.
2. Integrated stacks are a lame attempt for the hardware guys to stay relevant in a public cloud world. Public cloud providers are taking a top-down approach to infrastructure with most relying on generic commodity servers, storage, and networking rather than buying from the big-iron providers in the stack wars. This consolidation mirrors the dying breaths of the British auto industry, which attempted to retain their hold on the market through scale rather than competitive products.

It’s not immediately clear whether one, both, or neither of these arguments truly provides the impetus for these hardware stacks, but the vendors are all pushing them as cloud-ready. Certainly, the advent of cloud computing is a major factor.

Only the Beginning

This is only the beginning of the stack wars. I have asked the Gestalt IT authors and Tech Field Day delegates to weigh in on this subject, and expect to see a number of insightful pieces written on the topic in the coming weeks. I will continue covering the topic as well. Please consider subscribing to the Gestalt IT RSS feed or following @GestaltIT on Twitter to stay up to date on the latest developments. Thoughtful comments and responses are always welcome as well.

© Stephen Foskett for Gestalt IT, 2010. | The Stack Wars Have Begun!
Read more posts categorized as Cloud Computing, Exclusive, Featured, Networking, Server Virtualization, Storage, Top Story