1 September 2008

Virtualisation offers the opportunity to create infrastructures that provide more high-availability, flexibility and agility without using complex clusters

Günther Aust, Fujitsu Siemens Computers A group of the finance industry’s top experts answer questions on the development of virtualisation, and what it means for financial firms.

Virtualisation is today’s hot topic, and the latest developments in technology and business demands mean that it’s likely to see phenomenal growth. But this is not a new concept, having been around since the mainframe – so why are we now seeing renewed interest in it, from financial firms and vendors alike? Jacqui Griffiths asked the experts about the technology and trends behind virtualisation in the financial industry.

Why is virtualisation getting so much attention now?

Leslie Muller, DynamicOps: One reason is that a lot of the technologies are starting to mature. But really it’s been waiting in the wings for the past four years or so. A lot of companies, especially in the financial industry, have been looking for ways to become more dynamic in managing their data centres. As well as cost issues, their main concern is about being able to respond to market events much quicker, utilising their vast compute resources more efficiently. This move towards ‘data centre 2.0’ is completely underpinned by virtualisation technologies – not just hypervisors, but also virtualisation at the storage, network and operating system level. A lot of things have come together now, and it’s really starting to enable people to move toward that vision of the dynamic data centre. It’s just the right time to do that.

Ian Masters, Double Take Software: Companies are seeing that IT can deliver now, and they’re enabling themselves to offer additional or enhanced services. The underlying technology wasn’t necessarily there before. Now it is, and it’s giving people the ability to try something new.

Phil Dawson, Gartner: There are two levels of virtualisation. Many virtualisation projects are about consolidation of existing technology and optimising utilisation, and in that sense they’re retrospective and represent an IT benefit. Now, though, a growing proportion of virtualisation projects are about improving service levels, workload management and availability. These projects are about the virtualisation of new workloads, which is a very different business case.

Günther Aust, Fujitsu Siemens Computers: Today’s server systems offer an increasing amount of performance, and right now there is a clear trend: more companies are trying to move away from proprietary Unix systems, and they’re increasingly entering the open server or industry standard server market – Intel and AMD based systems.

The first motivation for virtualisation was to consolidate under-utilised servers and to do more with less. But virtualisation doesn’t just support consolidation – because of the hardware and software independence it creates, it also offers the opportunity to create infrastructures that provide more high-availability, flexibility and agility without using complex cluster systems.

Philippe Nicolas, Brocade: Virtualisation represents a cost killer enabler. When you virtualise applications, server, network or storage, you decouple the logical layer and the physical components – you make incompatible things compatible, with no downtime and no impact on business. This is a much better utilisation of resources and leverages existing assets, so the total cost of ownership (TCO) is improved.

Ian Warford, Microsoft: Virtual machine technology for timesharing on mainframes has been around since the 1960s, but it’s been pushed to the top of the business mind because these technologies now span the entire stack. Businesses of all kinds are facing challenging demands, and they need to be able to respond in a dynamic way. The expansion of virtualisation has come at the right time to answer those business needs.

Steve Yatko, Credit Suisse: People are recognising that virtualisation is no longer about a pure technology. It’s more about a paradigm shift in terms of data centre capability and business capability. The true value of virtualisation is far beyond server and desktop virtualisation – ultimately it’s the ability to layer on top of virtualisation, to deliver new capabilities that help achieve unprecedented advances in time to market.

How can capital markets firms benefit from virtualisation?

LM: Virtualisation allows you to do a lot of consolidation, driving up utilisation levels and lowering capital costs, or utilising the money you’ve put into your data centre much better. For many capital markets firms, the big issue is retasking – having virtual machines running different types of systems allows you to do risk calculations and in-depth analyses during the day while at night you may be using the same systems to do settlement or other back-office operations. It’s about being able to be a lot more adaptable.

We’ve seen the need for this in the past few months as events have unfolded in the financial industry. As some institutions start to struggle and it becomes public, their clients sometimes leave them and move to other financial firms, which then have to rapidly increase their capability around some product offerings. From one day to the next, they might literally have to increase their volumes three times or more, and you can’t do that with physical environments. Having your applications and systems run inside virtual machines allows you to grow a particular product or business unit overnight, to deal with the extra volume that accompanies market events. That’s been absolutely critical for some companies.

PN: We see many virtualisation technologies deployed for consolidation projects, as people realise that this improves their IT utilisation and reduces global cost for things like power consumption and operational expenditure. Financial companies have to react to a fast changing world and they need agility in both their business and the IT that sustains it.

In financial simulation, many firms use grid computing based on commodity hardware. The goal is to deliver fast results, and virtualisation allows easy and transparent provisioning without any impact on users or applications. Then, for specific usage, some departments select vertical applications on dedicated platforms, which cost more than traditional ones. The effect of virtualisation here is to standardise key core functions like resource management, data protection and business continuity by using the same horizontal open layer across server, storage and various IT units.

GA: Some key benefits that also relate to capital markets firms can be seen in Fujitsu Siemens Computers’ recent work for HypoVereinsbank in Munich. The company’s IT team started to evaluate virtualisation technology very early on. Initially, their main focus was consolidation, and we consolidated 650 physical servers in their data centre onto 43 Fujitsu Siemens blade systems. This saved the bank around 1.6m KwH (€180,000 in energy costs) per year in the data centre, and enabled it to stay in its existing data centre environment in Munich. That kind of saving is very significant to any business with high reliance on IT resources.

We’re still working with HypoVereinsbank, this time on desktop virtualisation in the data centre, and this is a key area where capital markets firms can benefit – we’ve had a lot of requests for application scenarios for brokers, who have to work with multiple systems at the same time. They normally have six to ten dedicated workstations under the desk and the same amount of screens above it, and they skip around the different applications for the information they need to take decisions. Their problem is not only the amount of physical workstations they need, but also that the data they deal with is very sensitive. They would like to store sensitive data and applications in a hosted data centre environment, but traditional approaches like terminal services are not allowed in that area due to legal restrictions. For those scenarios, financial firms are looking to things like desktop virtualisation to keep the individual systems separated in virtual machines, while consolidating and centralising them into a hosted data centre for security reasons.

Are there any potential issues raised by virtualisation in the capital markets? How are companies addressing these?

PD: With capital markets, as with any financial services company, you have IT-centric applications like internal IT processes and HR systems, and then the business IT services. Virtualising and consolidating the core IT is similar to any other business, but when you get down to the transactional and call centre systems, it’s very much down to the applications. If you’ve got independent software vendor (ISV) enterprise resource planning (ERP) systems, that’s like any other application, but if they’re in-house applications it can be much more complex.

IM: One big reason for virtualisation is to get more out of less, but from a workload point of view, some applications are really only going to deliver the right performance when they’re hardware hosted. It’s down to specific application needs – if you have a large application that will really only deliver the right performance on hosted hardware, then you need to think about whether virtualisation is suitable. But as long as virtualisation is managed correctly in terms of things like security, performance, and data protection, it can offer significant benefits.

IW: That’s one area where partnering with IT specialists can pay dividends – as virtualisation technologies continue to develop, financial firms need to know how to structure their virtual environment, and also to simplify its management to ensure compliance as well as agility. For capital markets firms, where successful trades can depend on split-second timing, it is crucial to ensure that resource-intensive business-critical applications are performing at optimal levels How important is the management of virtualised environments in this arena?

LM: In the past, you had one application tied to a server, and business units would not share those physical platforms. With virtualisation you share computer platforms and physical machines with multiple applications, sometimes across business units. It’s important to be aware of the end-to-end process, from the desktop right through to the back-end system and the market, and make sure you understand the impact.

For example, traditionally in a data centre we worry about how busy the CPU is, or how full the memory is. We really have to move beyond that and start understanding the service level agreements (SLAs) of that application or system. That might mean that the underlying environment is running at almost 100 per cent, but as long as your business system is meeting your SLAs to internal and external customers, there’s nothing wrong with that. The challenge is really that mind shift, for operations people and for the development community within the financial industry.

SY: Virtualisation is becoming more natively embedded in operating systems, enabling the ubiquity and pervasiveness that drive value across the IT landscape and deliver the beginnings of transparent capability. On top of that are layers of virtual management which can create highly available virtual environments that satisfy traditional, physical high availability as well as disaster recovery and business continuity plans – factors that are critical to capital markets organisations. The additional aspect is really about seamless access to the whole lifecycle of the virtual environment management layer. This is why Credit Suisse developed DynamicOps. We wanted to extend the value of our existing suppliers and technologies – to add value on top of that.

PD: Rationalisation can help reduce complexity and cost. If you look at a typical virtualisation process, taking a thousand workloads to a hundred servers on a ten-to-one ratio is the saving. But if you rationalise them down to 800 workloads on 80 servers, you’ll get even bigger savings. The process of balancing consolidation, rationalisation and virtualisation is absolutely key.

What benefits can Microsoft offer?

PN: Microsoft is playing a key role in virtualisation by leveraging Windows’ core functions to enable the dynamic data centre. It confirms that software is the key enabler for virtualised infrastructures that offer greater flexibility.

In many directions, such as servers with Hyper-V or storage with a file storage approach based on CIFS (the standard file sharing protocol for Windows) and distributed file system (DFS), Microsoft has set new de facto standards that help users define new and better IT usage. In fact, when people select Microsoft components in financial arenas, they continue to use these elements with recent Microsoft virtualisation technologies. It helps them maintain and augment the life of their IT resources, drastically influencing the TCO.

IM: Hyper-V is going to be a very valuable alternative for customers in the capital markets and others. Companies that already standardise on Microsoft will find it even more appealing from a support perspective, because they’re so familiar with the Microsoft operating system.

LM: When the Hyper-V becomes fully available later this year, it will attract significant interest. Microsoft has put significant amount of investment into this space, and I believe it will become a major player in the next couple of months. In developing DynamicOps, it was key that we utilised Microsoft technology, specifically its workflow and rules engine technologies. We worked with Microsoft from day one, and got a fantastic product out of it.

PD: At Gartner, we think virtualisation as a market is about six to seven per cent penetration of servers today, and that will grow to around 15 per cent by about 2010-2011. As it goes over 15 per cent that starts affecting the volumes of shipments for the systems – as systems get bigger, their value goes up, but units may go down. Vendors will compete more on the management and tools around virtualisation than on virtualisation itself.

GA: Microsoft has done an excellent job in expanding its management suite with a new strategic pillar, the System Center Virtual Machine Manager. It has been very clever in identifying the importance of management – it had a management product before it had an enterprise ready virtualisation product.

How do you think the industry will develop next?

PN: It’s all about partnerships and features integration. Users are waiting for global data centre solutions with application, server, network and storage to control and manage. They want to use commodity hardware with intelligent software, and that software will make the difference.

Standards are also important, as they contribute to reducing cost, and best practice legislation like ITIL is key for a broader adoption. One fact to remember is that IT is managed, controlled and seen with and through this virtualisation technology. The IT world is shifting from an application, data, server, storage or network centric world towards a virtualised centric approach where all the others are embedded.

PD: I think we’ll see sedimentation in terms of the hardware, and real differentiation will come from availability suites and systems management.

IM: In the future, we’ll see even more cores. There will be hardware developments in conjunction with virtualisation developments that allow even better use of the CPU in a machine, being able to thread it properly and having multi-core processors that can deliver performance on an almost individual virtual machine basis. We could also see modularised applications that can spin off a new virtual machine to add capacity as they hit a workload state or threshold, and then spin it down again when it isn’t required.

Obviously everyone’s use of a particular application is different, so it will be interesting to see if hardware and storage manufacturers continue to try and develop their products to fit in with virtualisation or whether virtualisation will have to fit in with the hardware developments that are occurring anyway.

GA: Fujitsu Siemens Computers’ dynamic data centre strategy is based on three main pillars: virtualisation, automation and integration. Virtualisation as we see it is not simply a concept; it’s a set of capabilities provided by that concept. Accordingly, while hypervisors are an important building block in the overall virtualisation strategy of a company, there are other things, like a uniform management of real and virtualized environments, to take into consideration. It’s becoming a much broader landscape.

LM: The data centre is going to require a very good management platform with workflow engines, rules engines, policy controls and so on, to enable it to deal with a very complex dynamic situation. This is the end state that DynamicOps has in mind. It’s a vision that Credit Suisse shares with Microsoft, and with a lot of financial companies. Microsoft understands what the data centre is going to look like, and is on the technology space, solving issues in critical areas of the real virtualisation space, from interfaces to hardware. Complementing that endeavour, companies like DynamicOps will offer solutions to deal with the complexity at a higher level.

Add a comment