There has been a spate of acquisitions in the cloud space, as companies are trying to fill holes in their service offerings. Recently we’ve seen NaviSite snatched up by Time Warner, CenturyLink buy Savvis, and OpSource go to DimensionData (owned by NTT).
What’s happening and what does it mean for cloud services?
The Cloud Landscape
The big players racing for cloud are Amazon, Microsoft, Google and VMware. Amazon got it all started with AWS. The platform allows customers to upload Xen virtual machine images and provides APIs to manage them. Microsoft is a .net-centric platform where you provide the code and Microsoft runs and manages it on Windows virtualized servers. With Google App Engine, you write Web apps (typically in Python), give the code to Google and they take care of the rest. VMware’s vCloud is a set of enterprise infrastructure and tools that allow service providers to compete with Amazon and offer services for the hybrid cloud – that is sliding applications from on-premise infrastructure to the public cloud and back again.
Each of these platforms has some type of storage approach. Amazon’s S3 is an object storage model. Microsoft offers storage for applications and SQL services while Google has the Python Datastore API, a schema-less object data store.
Then there’s OpenStack. OpenStack is open source initiative founded by Rackspace and NASA. It has grown a community of developers and contributors and is essentially a ‘Hail Mary’ against VMware and Amazon. OpenStack is far from being a production-hardened, five nines type offering and many have joked that it takes rocket scientists to run, a reference to NASA’s involvement. But OpenStack is open source, it’s free and it has momentum. The goal is to create a Linux-like or Lamp Stack-like community that can be a force in the industry and provide an alternative to proprietary offerings.
OpenStack has an object store called OpenStack Object Storage (codename ‘Swift’). It’s a very nice system according to developers for storing unstructured data and long term infrequently accessed information. It’s new and open source so there will be bugs that the community will have to fix over time. It’s important to note its limitations. It’s not a file system, objects must be less than 5GBs, there is limited flexibility (e.g. you can’t configure user storage quotas) it doesn’t support file locking and there are several other gaps in functionality. But like open source projects with momentum it will improve with time.
Enterprise Cloud Jockeying
In a 2009 paper entitled “Above the Clouds,” Berkeley electrical engineering and computer sciences whizzes provided an excellent overview of cloud computing at the time. Their definition of cloud explicitly excluded on-premise offerings and immediately sent the marketing pros at HP, IBM, EMC and VMware spinning to find a response. That pivot was so-called private and hybrid clouds. It was an un-coordinated but important initiative by all enterprise players to come to the defense of CIOs and on-premise IT infrastructures by offering the promise of simplicity, agility, velocity, better economics and a future vision of safety and control.
The most substantial effort in this space has been led by VMware. It’s vCloud offering and partner strategy is the gold standard for enterprise private and hybrid cloud offerings. Service providers such as CSC and Verizon have picked up on the trend and are beginning to aggressively offer cloud services for enterprise customers. These services directly attack the drawbacks of, for example, Amazon EC2 and S3 which include sub-standard SLAs (for many enterprises) and inflexibility accommodating certain auditing and security edicts of corporations.
The approach these service providers are taking typically includes a VMware + Storage Company + Networking + Service Provider solution. More often than not, the storage offering is EMC or NetApp with Cisco as a networking partner. EMC is aggressively marketing Atmos as an object store alternative to Amazon’s S3. While NetApp has purchased Bycast as its object play, it has yet to mainstream the solution for enterprise clouds. It’s important to note that in the case of EMC Vblock/Atmos and NetApp, these are enterprise-hardened solutions that the likes of CSC and its enterprise customers have grown to expect and command a price premium for higher levels of service. The key consideration for customers is not only cost, it’s the rental flexibility speed to deploy these solutions.
Meanwhile, Citrix, which is dominating the desktop virtualiztion space, has been trailing behind VMware in the enterprise cloud operating system market. It’s free Xen Server solution is widely installed in public clouds but is not a cash cow for the company as is VMware’s vSphere. The move to buy Cloud.com will be viewed by some as either a brilliant strategic flanking maneuver or an act of desperation. Personally I think it’s both.
Cloud.com’s primary business is cloud computing, not cloud storage, and as such can be used by Citrix as software to sell on top of XenServer, in a bid to make it more competitive with VMware. VMware has the dominant share for cloud operating systems in the cloud compute space and this move by Citrix is a strong message that it is throwing down the gauntlet and really staking a position in this market. Whether the cost was $250M or $500M, it still doesn’t mean that Cloud.com software is enterprise-ready, but Citrix is clearly putting its enterprise weight behind the OpenStack initiative to help get it there. We’ll see how this acquisition pays off relative to its market position versus VMware and its new vSphere 5.0 product in the coming quarters.
The Enterprise Whales
The large enterprise IT vendors—Oracle, IBM, EMC, HP, NetApp—are not going the OpenStack route because the bulk of their revenue is derived from conservative, Fortune 500 companies who expect battle proven IT solutions, not open source initiatives. While some of these firms may support OpenStack over time, they need a margin offering that allows them to make profits.
For these large IT vendors, their current strategy has been to maintain their position that they are the infrastructure “outfitters” to the cloud—their biggest challenge remains the increasing number of their customers shifting to the pay by the drink models brought mainstream by Google, Amazon and Microsoft Azure, among others. And as we’ve highlighted before, these big cloud service providers are not using IT gear from the big OEMs. They’re using high-density, commodity servers and storage systems with their own secret software IP layered on top. So the question becomes, how do they participate in a growth market and avoid market share losses to the likes of Amazon—which did $500M in revenue last year and is expected to do another $750M this year—and offer a enterprise-cloud storage and compute service of their own?
In a recent article explaining the executive succession plan at EMC Chris Mellor of The Register pointed out that “Should EMC offer its own cloud storage service for enterprises? That would be the ultimate hybrid cloud, wouldn’t it? EMC storage in the private cloud and EMC storage and service being offered as a public cloud. Perhaps EMC should consider buying its way in to this market; Nirvanix looks a tasty morsel.”
EMC has been adamant that it won’t compete with service providers and Atmos seems to be its preferred solution – so I’m not sure I buy Chris’ Nirvanix scenario. The more likely scenario is put forth by another writer.
Tony Asaro of SearchStorage recently noted that “I can see enterprise cloud storage playing out with EMC driving Atmos within its own customer base…EMC will get customers and generate business, things it tends to do very well. Expect a major IT vendor to acquire Nirvanix and then go head-to-head with EMC Atmos. Those two vendors will dominate this market, with other vendors having pockets of success here and there but never assuming a dominant position.”
Asaro’s point is more logical to me. Companies like Oracle, IBM and HP in particular, and Dell to a lesser extent, need to have an offering to compete in cloud storage. IBM and HP in with large global services organizations are falling behind companies like CSC and other service providers in the cloud storage space. Service providers are motivated to partner with VMware and EMC and there’s significant momentum behind their cloud offerings.
The focus on Nirvanix as an acquisition candidate is interesting to me. The company was dead quiet for a while but has been making significant inroads in the enterprise cloud storage market—gaining traction in hybrid clouds faster than traditional players, poaching senior talent from Google and racking up some wins in the healthcare and entertainment sectors. While others are doing object storage Nirvanix is singularly focused on the enterprise buyer and may be a target for whales trying to compete with the VCE/service provider strategy and offer an alternative to up-front CAPEX and high maintenance cost storage. IBM and HP with captive services organizations are the companies that have the most to gain by acquiring an object store with a scalable file system that can act as the hybrid cloud data store. Because Nirvanix owns and operates its own public cloud network the acquiring company has to have an appetite for services, as do HP and IBM. Another play could be Rackspace or other service provider picking up a Nirvanix to harden their cloud storage offerings.
While there’s been a lot of action by VC’s funding the cloud gateway companies (e.g. StorSimple, TwinStrata, Nasuni, etc. the freeway onramps to the cloud—are short-term products. Tier one OEMs will increasingly incorporate cloud gateway functionality in their storage systems via high speed Ethernet connections embedded in their boxes – e.g. an EMC VNX or NetApp filer with a built-in gateway to the cloud. As such the market for a separate appliance to upload data into the cloud of your choice and store some data locally on it will be pressured.
The greater value in my view will be on providing services to store data in the cloud long-term—for backup, archiving, images and other emerging use cases. Look for all the major enterprise players to have some type of offering (either directly or through partnerships) by the end of 2011.