Some posts ago I stated that Personal Cloud market is immature: in fact, as although I suggested it in my last post ”the cloud industry is still in its infancy”, and in that analysis found objectively that the Cloud market has a lot of features that show it. However, for example, a very recent fact that add a new point to this “maturity” controversy is that Amazon signed a few days ago a very significant and important deal with the Central Intelligence Agency (CIA) to build the agency a private cloud (a 10-year contract valued at $600 million).
So, let me say that general cloud market is not mature, but I couldn’t say immature (please forgive me for this easy words game): disregarding at the moment other features of this market (as its great dynamism, players are coming from different IT areas, prices policy not easy to compare, different performances, weak SLAs, etc.) I agree that some players are in this market since long time, and one of them, Amazon, with its Amazon Web Services (AWS), holds about 70 percent of the IaaS market (depending on the different reports and analysis) and it’s both the big innovator in this market and its big driver (making it affordable, accessible, and broadly relevant to the current IT market, because it has a well built cloud services catalogue that it’s continuously widen and making benefit from it), but so far (or until very recently) some big IT players as HP, IBM, Oracle and others are outside (or almost) of this market, so mergers, acquisitions and consolidation have been expected to happen not only on Service Providers (as we’ll see further in this post) but also in solution vendors (e.g. a few mints ago VmWare acquires Nicira to adds cloud networking functionalities to its platform as well as very recently acquires Virsto too, to add an object storage solution to its emerging cloud platform built it from its impressive virtualization solution).
Yes, I mention specially those 3 giants (HP, IBM and Oracle) because recently they move and step into this market with a very different policy they have done until then, and I’m going to lightly analyze their last movements:
- Some months ago, after to have been trying to develop its own solution, HP embrace OpenStack and announced that its “HP Cloud” was built on OpenStack (good for an OpenStack supporter as I am), which could be attractive for enterprise managers worried about being locked into a Cloud Provider or Vendor, but be careful about that: i.e., since HP uses an real Open Platform it bets and offers openness and portability, but since they offer proprietary extras you could fall partially captive of its approach:
- HP Cloud IaaS services include, for example
- “HP Cloud Object Storage”, i.e. HP’s version of Amazon S3 ($0.12 per gigabyte).
- Also, of course, HP includes compute resources (the smallest offering delivers 1GB of RAM for $0.04 per hour, cheaper than RackSpace that charges $0.06 per hour for a 1GB machine).
- HP also offers a CDN (Content Delivery Network) Services (built into this object store, but at $0.12 per gigabyte)
- and “HP Cloud Block Storage Services”.
- According to Gartner its SLA (Service Level Agreements) are quite bad, only Amazon have worst SLA, but I don’t doubt the former will improve them as the time (and service) goes by, and the latter when needed (forced by competition, when some company really challenges its market prevalence).
- Finally, besides IaaS service, HP also offers PaaS services or border line services as “MySQL as a Service” or a plug-in for moving Drupal installations to its IaaS, and so on.
- HP Cloud IaaS services include, for example
- A couple of weeks ago, IBM announced that all of its cloud offerings would be built around OpenStack, once again a great for the fans of this open source cloud platform. According to IBM, they will not offer its own packaged version of the OpenStack (as RackSpace, StackOps or Red Hat are doing), but to include OpenStack in their approaches.
- In one hand, unlike RackSpace or HP (mentioning only examples companies that use OpenStack), IBM doesn’t have a Public Cloud yet although is published in its web site, there’s not self-service model (as required by NIST to see Cloud Computing): you have to contact and wait for the phone back to you.
- In the other hand, so far IBM has Private Cloud software approach, SmartCloud Foundation, which offered (according to IBM) much the same functionality as OpenStack and they donated functionality from its “SmartCloud Foundation” product to the Folsom release of OpenStack (indeed IBM is the third big code contributor to OpenStack Folsom release), however from now on all future versions of SmartCloud Foundation will be built around OpenStack, but adding features as management, security, orchestration, and more: in fact, the first product to bundle OpenStack is “IBM SmartCloud Orchestrator”, now in beta testing phase, which enables customers to compose cloud services using a drag-and-drop interface. The Private loud seems to me the natural market at the moment for IBM; in fact IBM claims that it had 5,000 private cloud customers as of last year.
- Taking advantage of OpenStack interoperability, IBM working toward making workloads portable from cloud to cloud, avoiding cloud lock-in (perhaps the primary inhibitor to public cloud adoption, after security, as already stated in a previous post). By the way, OpenStack includes in its next version, “Grizzly”, scheduled for 4 April 2013, an incubator module called “Heat”, that will provide a Basic Cloud Orchestration & Service Definition, i.e. to orchestrate multiple composite cloud applications via a template format as AWS CloudFormation does. But IBM plans going quite farther even, and it’s fostering an emerging Oasis standard known as Topology and Orchestration Specification for Cloud Applications (TOSCA): a way of describing a workload application that’s independent of the actual infrastructure that’s running it.
Next post will focus in the Oracle’s changing and maybe misunderstood Cloud position (Oracle and its CEO Larry Ellison used to disdain Cloud Computing, e.g. he called cloud computing “nonsense” in 2009), that in some way reminds me the IBM’s one when its mainframes ran mission-critical applications on legacy databases, and a new (by then) generation of infrastructure vendors (DEC, HP, Sun, Microsoft and Oracle) challenged it and disrupted the old IBM model: it was conflicted about selling the lower-priced, lower-margin servers needed to run them.