Content Copyright © 2009 Bloor. All Rights Reserved.
This blog was originally posted under: The Norfolk Punt
Cloud computing, as Bloor’s Cloud specialist, Martin Banks, keeps telling me, means may things to many people. In fact, the meaning attached to “cloud” can change radically during a conversation!
So I jumped at the chance to hear what “The Cloud” means to IBM – since IBM has been doing bureau computing and virtualisation (which, seem to me, to encompass at least some of the Cloud ideas) with mainframes for years. The event was held at IBM’s first Cloud Computing Centre outside of North America, its Dublin Innovation Centre.
Erich Clementi (Vice President, Strategy and GM, Enterprise Initiatives, IBM Corporation] set the stage. He sees Cloud as a different, but evolutionary, way of managing IT – and IBM is concentrating on the datacentre savings it offers from efficient utilisation of shared resources and the profound implications of self service. The Cloud model has similarities to the Amazon shopping model – the datacentre customer (in the wide sense) does all the work in choosing the provisioning they want and just presses a button to make it happen, using clever automation. Perhaps the banking ATM offerings provide an insight to what Cloud means too – ATMs drove standardisation for a whole industry.
You can sum up the value message for Cloud as (in my words, more or less): “It addresses the issue that you never have enough computing power when you really need it – but that most of the time you’re paying for more capacity than you’re using”.
Nevertheless, Clementi points out that the cloud isn’t a universal solution and it won’t suit some workloads (the migration costs and risks may be too high for conventional, mission-critical, database transaction processing and ERP workloads, he says). He also points out that, while Enterprise management is happy with Cloud security based on a Virtual LAN, it wants more monitoring and hands-on management of what people are actually doing than Clouds, which often come out of Academic environments, often provide today.
However, some services could become standardised for Cloud deployment: Web infrastructure applications, collaborative applications, development and test environments (this seems to be IBM’s key Cloud delivery at the moment) and High Performance Computing services.
And some applications may only be truly feasible on the Cloud: high volume, low cost analytics services; collaborative business networks and industry-scale “smart” applications.
Other IBM executives gave us a real insight into the way IBM sees Cloud computing arising, in part at least, out of its past mainframe bureau computing culture. Timothy Jones (Manager Information Services, Zurich Research Lab, IBM Research), for example, talked about the use of the Cloud in IBM Research, a particularly fertile test bed for cloud initiatives. He talked about virtualisation and datacentre consolidation and the changing role of the CIO as a precursors to the adoption of Cloud and identified three themes driving CIO strategy that are also relevant to Cloud adoption:
● decoupling business processes and data from the underlying IT applications;
● decoupling business applications from the infrastructure;
● establishing new collaboration environments that support business innovation
An interesting “Use Case” came from Robert E. Rosier, CEO of iTricity/GlidePath, a Virtual Datacenter Operator & Capacity Provider. It currently hosts more than 500 x86 cores, more than 50 P-series CPUs and over 200 Tb of data in have 5 datacenters, all connected via a fiber based backbone. It bills on a subscription and/or Pay per-Use basis (depending on the complexity of a user’s requirements) and although Cloud is only a minority percentage of its total business, it is growing twice as fast as conventional hosting.
Most interesting is the “compliance” value-add of the iTricity offering: clouds can be supplied that meet the infrastructure management requirements for SOX, BASEL II compliance etc. – although that is not all that’s needed for compliance, obviously; and I’m not quite sure in my mind exactly who the regulator comes after if you fail compliance. A key point here is that using the Cloud doesn’t necessarily mean that you don’t know where your data is – if you want to pay more for it, for example, you can specify an SLA with iTricity that means that your data all resides on a dedicated facility inside the Cloud that is reserved just for you.
Speculation on possible Cloud futures is always interesting. However, they are not entirely castles in the air. The NESSI (Networked European Software & Services Initiative) Reservoir project, for instance, is an EU-funded research project focussing on developing a next-generation Service-Oriented Infrastructure (SOI), based on Cloud Computing and the integration of grid, virtualisation and service management technologies. It is going to develop an “internet-scale data centre” (read the presentation to the NESSI General Assembly).
One possibility is for future “industry clouds” offering industry-specific collaborative value-adds. These could support, for example, things like the CLS (Continuous Link Settlement) infrastructure which mitigated counter-party risk in the financial markets even while Lehman Bros. was failing.
Another possibility is an “HPC on demand “ Cloud, providing High Performance Computing (HPC) for low-cost data-intensive analytics. This could address a very real issue – with very large datasets, data cleansing and sampling may be an overwhelming part of the latency and other overheads which can make “near real time” decision support applications infeasible. A technology such as Pervasive’s DataRush could be deployed on a flexible multicore infrastructure as a Cloud service and could be accessible to, for example, federations of small shopkeepers, giving them customer analysis facilities to rival the big supermarkets.
And, it seems likely that the Cloud will be a hybrid continuum between different private and public Clouds, in practice. However, one particularly interesting speculation for me, is that it might finally bring about the end of the physical IT group. We can now really integrate IT with the business (as required by ITIL v3) instead of just aligning it with the business: IT professionals can work in the business units they support, using Cloud self-service provisioning to set up their own development environments. These could be co-ordinated using Cloud-based collaboration to instantiate a virtual IT community facility.
Nevertheless, no matter how attractive Cloud futures appear to be, always remember that history suggests that new computer-based technologies add to old technologies, they don’t replace them. Which means that we can expect the Cloud to co-exist with conventional technologies for many years.