1 November 2003

To have truly efficient IT systems, organisations need to take stock of their current IT strategies and consider new ways of working. John Liddle examines ways of improving operational efficiency in enterprise organisations.

It is a period of belt-tightening across the global economy. The skyward markets of the 1990s have long-since gone, but most markets have just about managed to beat off full-blown recession.

Part of this has been due to a change of outlook. While there is a continuing need to improve sales and achieve targets, there now needs to be a further emphasis on internal processes. In an effort to stave off the effects of worldwide economic slowdown, businesses have been looking inwards at themselves and how they can make their internal operations more efficient and cost effective.

"Operational efficiency is about looking at business processes and finding how you can improve them to get things done faster or for less cost," says Nick Toozs-Hobson, ISV relationship manager EMEA at Hitachi Data Systems. "In other words, your output over input ratio improves."

While it has become natural to turn to technology to seek solutions for all of our problems, it is important to conduct a proper analysis to see exactly what has to be done. It is highly unlikely that technology alone will prove to be a saviour.

"I think that for improving operational efficiency in any organisation, it is important to understand where your organisation is," says Ofer Plaut, manager of professional services at Getronics Israel. "You need to conduct an analysis - you cannot just decide that installing 'Product xyz' will improve your operational efficiency."

Hans Sparkes, strategic marketing program manager for server consolidation at Unisys, says that when they are asked to look at a client's system they are able to take a lot of the risk out of a big project because they can guarantee how much it will cost at the initial assessment.

"We can do a very thorough assessment job up front, which deals with a lot of the concerns of the client and takes out the risk, because we can say in the design how it will be," he says. "At the end of an assessment we will deliver a fixed cost proposal because we can identify all of the characteristics of the applications through tools that we have developed ourselves because we understand the underlying components of the operating system. There are no cost overruns so it's a great way of de-risking a consolidation project. The client knows exactly what he is in for right from the beginning."

Getronics is engaged with Israeli banking giant Leumi in helping them upgrade from Windows NT 4 and Exchange 5.5 to Windows 2000 and Exchange 2000. The initial assessment showed that they needed far more than a technology solution.

Plaut says: "The first and fundamental thing we do is follow our own methodology to improve our product delivery. We analysed their situation and before even speaking about technology we discovered that they were not ready to provide operations management for the directory by themselves. So the first thing we need is to understand their situation at the time and establish a group that manages the directory. Now, the emphasis was on establishing a team with very clear ideas on how to manage the operation, because if you don't know what you are supposed to do and how to do it, and you don't have the tools to monitor yourself, then you can never improve."

Following the initial assessment, Plaut says that they then go on to leverage anything that doesn't cost any additional money. "From a technical point of view, for the daily operation we always use the built-in tools to manage operations more effectively and efficiently. So we use an extensive set of integrated tools within Windows Server 2003 to manage the Active Directory and to manage the servers. There are about 260 servers for Leumi in Israel and 40 worldwide, so it's quite a large operation to be able to manage all these things."

Cap Gemini Ernst & Young (CGE&Y) is working with Microsoft and Intel to offer a joint solution to help clients upgrade to SAP R/3 Enterprise to reduce the total cost of ownership. Martin Dunn, global ISV solution manager at CGE&Y, says: "Over the next couple of years, more than half of SAP clients will need to upgrade their current version, so this offering encompasses the SAP upgrade and platform migration to the Microsoft operating system and SQL database."

Like with Unisys and Getronics, they go to the client first to do an assessment to see if the upgrade makes sense for their business and IT needs. "That's a free assessment and then we will do a more detailed assessment over four-to-six weeks, where we take a more detailed look at their business processes. Then we'll come back with a business case where we recommend how they go forward," he says.

"We divide it into three areas. The technical upgrade focuses strictly on their SAP environment and will deal with platform migration. A functionality upgrade looks at core business processes and takes advantage of some of the new functionality of SAP. The third type is a strategic business improvement upgrade, and that is for customers who are looking to add to their CRM or business intelligence solutions. This is the biggest type for us, this is where we think we can make the biggest impact."

Dunn says that because the Microsoft platform is cheaper than so many of its competitors', while retaining scalability and reliability, there is no reason for customers not to migrate. "Not given the significant price difference," he says.

The biggest way that IT departments can improve the operational efficiency of the business - and make an active contribution to turnover - is through server consolidation. Toozs-Hobson says that because Microsoft applications are by their nature so easy to deploy and highly scalable, it has become a victim of its own success. "One of the biggest challenges Microsoft has faced is that as they become bigger and more successful, and they have built applications that are of more use to help people improve the way that they work, operational efficiency has suffered. Because to grow all you had to do was add another server, but the consequence of that is that it becomes more difficult to administer because you have more nodes on a network, more physical servers to back-up every night, and so it builds in process inefficiency."

Unisys' Hans Sparks says: "It used to be the case where you had one application per server if you were lucky - often it was one application to many servers. We've got to a stage now where people are saying that managing it all is too hard, there is too much cost and too much hassle."

He says that at one event he recently attended, a survey of CIOs showed that 29 per cent of them used more than 1,000 Windows servers, with 58 per cent using more than 250. "At that sort of level, you are getting into some of the difficulties of management, control and disaster recovery," he says. "You can imagine with the mixture of NT, 2000 and 2003, the patch control alone must be horrendous. The sheer number is the biggest issue and the cost just escalates."

Toozs-Hobson says that there is no greater example of this than with Exchange. He says that nearly everyone uses the application for its e-mail, calendar and address book functions, but once you get to more than about 500 users on a single server, you have to add another. So they have worked with Microsoft to find solutions to improve operational efficiency.

"We worked with the United States Support Lab in Microsoft to find a way where you could run 5,000-6,000 users on a single server," he says. "So instead of having 20 or 30 servers, you now have two or three. You can create back-ups on the same device without stopping Exchange from running, and you can restore it on the same device if there is a corruption or a virus."

The consolidation also means that IT departments can go from perhaps five or so administrators to just one. Toozs-Hobson says that they can perform a similar operation with SQL Server.

"The next step we are working on with Microsoft is to handle the proliferation of file and print servers - which again have the same problems as Exchange but even more amplified," he says. "Network Attached Storage (NAS) is a file and print server that doesn't have any disk and Windows Powered NAS is a slimmed-down version of the software that allows file and print to go through a specialist server. They allow huge amounts of consolidation."

But if you are planning to upgrade your systems and consolidate your servers, it can make sense to tackle both projects at the same time. With 58 per cent of Exchange users still on Exchange 5.5, EMC are providing solutions and servers to help them migrate to Exchange 2000 and to consolidate.

"Once they get there we try to get them to go with a scale up solution rather than scale out. So instead of having 500 users on a single server, we are getting customers up to 5,000 users per server," says Todd Donaldson, director of Microsoft solutions at EMC. "In Exchange 2000, they designed the database so that you could chop it up into 20 little pieces for back-up and management purposes and so it's almost like putting 20 databases on a single server. With that there is a lot less administrative cost."

EMC has worked with Microsoft's Exchange team to develop automated back-up capabilities that integrate with Exchange 2000 Server APIs to make it more storage-aware. EMC Clariion and EMC Symmetrix storage platforms are "build-as-you-grow" solutions.

Donaldson says: "We certainly suggest that if you are going to go through migration - which is a fairly significant project for most people - then get your infrastructure right. Don't perpetuate the legacy of scale out and all the mess that goes with it. Some people just want to do one thing at a time, but there is a substantial amount that say, let's kill two birds with one stone."

EMC worked with Unisys to implement a solution for IT-Austria, one of the largest application service providers in the European financial services market.

IT-Austria deployed two 16-processor Unisys ES7000 servers running Microsoft Windows 2000 Datacenter Server, Exchange Server 2000 and Microsoft Cluster Service. The ES7000 servers are linked to an EMC Symmetrix-based storage area network that uses EMC GeoSpan software to enable automated failover for disaster recovery and constant uptime.

Unisys is also implementing its own server consolidation project and is making good use of its ES7000 servers. Unisys CIO John Carrow says: "We are a reflection of our customers in many ways and we've made it a point to use what we sell. So we think server consolidation is very important because we've done it ourselves, we see the benefits of it and we think we've got good hardware to allow it to happen. With the ES7000, you don't find any bigger or better and it's at the right price to allow this to happen."

Hans Sparkes says that companies can expect 40 per cent reduction in infrastructure costs through consolidation of platform and application services. "We've found from clients that 40 per cent reduction over a four year period is achievable, and that is purely because you take out so many of the applications and the associated costs because many of the costs are hidden," he says. "People don't appreciate that it is not just the floor space - there is the support, maintenance and software. These things may appear small for each item but when you add them together it is substantial. So even if you could get the hardware for free, there is still a high total cost of ownership. We're looking to make a big cut into that, not just trimming it."

But Unisys' approach to server consolidation is about so much more than just the ES7000. Hans Sparkes says that they make sure that clients understand that server consolidation is not just about getting rid of boxes - it is about a change of mindset that affects how the entire IT department operates.

He says: "Obviously there are different approaches to consolidation - one is just putting in more powerful boxes running the same application, and that is a way of squeezing more users onto an application. But that still gives you small islands of information and those islands will still grow in their own right so you are not making the most efficient use because you are not balancing the workload. There is still scope for improvement because you still have a serious number of servers - following this method, you may only have gone from 1,000 down to 700. The next logical step is to put mixed applications and workloads together. Then you can balance the workload so you are delivering more work from fewer resources."

The Unisys approach is to go for a "big systems" attitude with mixed workload as the primary focus. With this in mind, it is not uncommon to consolidate 50 servers into one, although it depends on a number of factors including the age of the servers and utilisation. "At the end of it we will give the optimum configuration to supply that workload and we will build capacity within for future growth."

Because of the size of server consolidation projects, Unisys typically breaks it up into two or three phases. "You need to make sure you are getting it right and gain confidence that you are making a return," says Sparkes, adding that the clients usually see a return on investment inside six months for the first phase. "It is not feasible to do a whole system in one go, and we wouldn't expect a client to do that, because these are business critical systems. You need to take it one group at a time."

Unisys will take a selection of servers they think should be consolidated first, for whatever basis, whether age or it is a logical group of applications. "You need to see if the target system is delivering as expected before you turn the old stuff off," says Sparkes.

What you have at the end of this is the ability to deliver to lower cost using industry-standard Microsoft and Intel platforms. Security is also improved because there are fewer points of access.

But once the new boxes are in, Unisys needs to ensure that IT staff are in the mindset of managing fewer servers more tightly. "Consolidation isn't just one project, it's a way of life," says Sparkes. "It's a way of managing the infrastructure differently. Rather than saying, 'I need a new application so I'll buy a new box,' the request is, 'I need a new application or more capacity, where am I going to put it?' It becomes much more about service delivery."

This is where IT departments can at last be seen as drivers of business revenue and not just a cost. Sparkes says: "You are doing forward-looking capacity planning, which requires a tighter relationship with the business. That is good, because you are seeing what the drivers are and how IT can help deliver competitive advantages, rather than just managing a box."

This requires a complete change in how the IT team is managed and usually requires Microsoft and mainframe teams to be mixed in some way, which Unisys can help with. "It is a culture change," he says. "You've got to change a way of life that has been going on for many years. But part of our delivery vehicle deals with the transfer of information, including the operational side of it."

In an odd way, the current drive to consolidation couldn't have come at a better time. The proliferation of servers means that the increasing complexity of handling them would have led to consolidation anyway. The more recent focus on security and disaster recovery is adding to the magnitude. The fact that CFOs are tightening their belts to squeeze through the current economic impasse makes server consolidation even more timely.

Add a comment