Come together

Speaking exclusively with Business Management, Doug Gourlay, Senior Director of Marketing and Product Management for the Data Center at Cisco Systems, assesses the long-term benefits of improving data center efficiency through consolidation and virtualization, and walks us through some of technologies enabling this in enterprises today.

When asked to list the most pressing problems and challenges they face today, the reply from our customers is almost always the same – first and foremost it is data center consolidation. Hewlett Packard is one such example, and their case illustrates why getting it right is important. The company recently met with Cisco to discuss the consolidation of their 80 data centers worldwide to just six in the US. Accordingly, we planned the design of a network to support these six data centers, one that would be capable of supporting a global network of tens of thousands of users. When the company publicly announced this planned consolidation four weeks later, their stock price rose four percent, because the changes meant projected operational cost savings of around one billion dollars.

Those facing consolidation want to remove the negative side of distributed, workgroup and departmental computing that they have experienced in the last few decades, where servers were moved from the mainframe to servers and client server applications. Whether driven by the network, the server team or the application, servers started appearing under desks, in wiring closets and broom closets. Through data center consolidation, centralized IT management aim to establish a strong IT governance model. Fortunately, the technologies exist today to overcome the organizational challenges associated with this and, most importantly, to ensure that the user experience isn’t degraded.

Today’s data centers – the challenges

Over the last 5-10 years, data center complexity has risen – storage and storage networking has been introduced, as have IT network fabrics. Furthermore, in the last 13 years, computation capacity has increased by four orders of magnitude. All of this means that enterprises today must manage a vastly more complex environment than they faced a decade ago.

As a result, the operational expense and people assets necessary to support and manage this infrastructure is now much greater than the telecommunication costs of interconnecting a distributed environment. It has become increasingly cost-effective, both from a telco perspective and a people perspective (and also in terms of utilization of assets), to bring servers back centrally. Few of us would be happy with an employee who worked 20 percent of the time, so why be content with a server that does so?

A second key data center challenge for enterprises is business continuance and disaster recovery requirement. We’re finding that, even down to the commercial level, there is now a strong focus among companies on insourcing or outsourcing disaster recovery models.

Disaster, whether man-made or natural, can seriously impact your business. When Hurricane Katrina hit, many companies found that they lost their data centers and, without a functioning backup facility, business ground to a halt. One solution is to use an active-active or active-standby redundant facility, ensuring that those facilities are far enough away to be protected against the power outage, natural disaster, etc.

When the data center works right, it can protect against an outage in such a way that the end-user doesn’t even know there has been a problem. There is a collection of dedicated technologies and applications to enable that.

The last thing that is becoming a challenge in data centers – which is possibly exacerbated by data center consolidation – is power and cooling. Where we once had three or four servers in a rack, we now have 20-40; the density has increased dramatically. Organizations suddenly find themselves faced with 12-15,000 watts per rack and next-generation blade servers that project 30,000 watts, all of which needs liquid cooling. Businesses are running out of capacity – not only space, but power.

So why consolidate?

The foremost benefit of consolidation is a reduction in operations costs. It is very hard to refute the fact that, in Hewlett Packard’s case for example, six facilities are far easier to manage than 80. There’s also the cost of running the facilities themselves – cost of power, interconnectivity from telecom links, etc. It’s all about economy of scale. Enterprises can often do the same work with fewer, larger and better-interconnected servers. Bringing servers together in a central facility is one example of how consolidating data centers can have a direct impact on the cost of operations or the future capital outlay required.

Storage is another good example; moving toward a single consolidated storage area network allows the enterprise to achieve more efficient utilization of the assets they already own.

There are also advantages in improving business productivity. As an example, when we as a company consolidated multiple storage area networks to a single large and global storage area network, the number of people needed to support our storage reduced from 32 to eight. These people were then moved to other areas of IT that needed their expertise and were retrained, enabling us to make better use of our assets.

One final advantage that is often overlooked, and that we achieved through our storage consolidation, is an improvement in service velocity. Previously, if I needed a terabyte of storage for a new application, IT would tell me to wait a month. Today, they can have it online for me tomorrow. We also reduced the total cost of ownership of managing that storage by nearly 80 percent and achieved more efficient utilization of our capital assets.

The data center of the future

In the next year, we will see an increasing focus on provisioning. We will start to enable provisioning in the next 12 months, which will generate a very dramatic shift in enterprise capability in the data center.

A key aim for us is finding ways to make the network more usable, and moving towards a network that is more adaptable to the business needs. The aim is to be able to install a server, plug it in, turn it on and walk away. That server will then connect to the storage network and the local area network, it will be mounted to the drive it needs, will load the necessary applications, and be on the right LAN segment, network services, firewalls, intrusion detection, application networking, etc. to function as a useful part of the overall IT infrastructure of that enterprise.

The other trend that will continue to grow and that will absolutely change the way that data centers are designed and operated is server virtualization technologies.

Virtualization and the data center

Virtualization is probably the most overused term in networking today and everyone’s experience and definition of it is different. Virtualization is about taking a resource and either extending its reach out to groups that weren’t able to access it before or segmenting it in such a way that one server looks like 10. The most common example of virtualization, and one that is receiving a lot of attention today, is server virtualization. Server utilization is often less than 15-20 percent, and this means that a huge opportunity exists to reclaim those servers, use them more efficiently, or to add multiple virtual machines to them. Other forms of virtualization that are happening today, but are perhaps receiving less airtime, include storage virtualization, storage network virtualization and local area network virtualization.

To be relevant to CIOs today, we need to consider virtualization as a key technology area. At Cisco, we moved our server load balancing and content switching, and application front-end and SSL technologies into the virtualization realm with the recent introduction of the Ace module and its customer contacts. Those customer contacts can be applied to some of the business challenges of compliance – for example, segmenting the customer-facing applications for traceability – and we can do that with less equipment than our competitors. And, of course, less equipment also means lower power and cooling costs, and less space required. The ROI can be measured in months – weeks, in some cases.

In every major product category, we’re shifting from a focus on consolidation alone – which is about scale, availability and performance – to a focus also on virtualization.

Demonstrating real value

We have a very intimate and direct relationship with our largest customers and take the time to educate them, not only about what the products do, but also what business results they can achieve through them. We also encourage our own IT department to interface with customers, to communicate what problems they are experiencing with products, and how they can be overcome.

There will continue to be ongoing outreach to the analyst, consulting and press communities, and we will continue to attend the key industry events to build awareness for Cisco.

We also intend to build our partner programs, because our partners may have a better or more diverse relationship with customers, or a broader, in-depth focus than we do in certain areas. At Cisco, we aim to communicate more strongly the strategy and vision of the data center and to quantify some of the diverse business benefits of consolidation and virtualization technologies – the benefits we’ve experienced in-house, and those seen by our customers.