How Critical Is The Data Center Operator to End-to-End Performance Management?

How Critical is the Data Center Operator?

As I’ve mentioned earlier, one of the unique features of this initiative is that we are sharing with the community at large key issues and the most interesting discussions that ensue (in the build-up to the panel) between the eco-system players and the various carriers involved in this initiative.

This has several benefits, which we’ve already articulated here, for instance, and here.

The chief among them is to give the community valuable insights that emerge in these roundtables and that are invariably lost by the time the final panel comes around :-) , or at least not one’s we’ve seen shared with the community in quite this way before. (A non-trivial side-benefit is that it provides a living list of relevant issues, a documentary of the evolving though processes and many teachings, all of which get documented for future use by the community, in a concise and logical manner.)

It also gives the community the valuable opportunity to be involved in dialog and discussion leading up to the panel, providing a much richer set of issues for contemplation, and enabling each attendee (and panelist!) to derive a whole lot more from the discourse. Plus, it involves the telecom community in the conversation, and enriches the panel with the collective intelligence of the community.

In this post (and several to follow), we will summarize for you a number of important issues that were (and are being) discussed in our Roundtables, and logically show you what the evolution of our thought process has been, as we pondered this subject, teasing out the issues so we could to select the most relevant ones for discussion in the live panel.

So, let me start with a very interesting subject and one of the first that we discussed, namely, what is the role of the data center operator in end-to-end performance management today?

Turns out that in considering the operator landscape for performance monitoring and management, we had initially thought of only four operator types: wireline, wireless, cable/MSO, and over-the-top (OTT), and laid down some of the performance management requirements for each of them. (These emerged from the many discussions that I had over the last few months with colleagues from each of these operator segments (and many eco-system players), and from analyzing those and combining them with our own insights into the subject.)

However, both Ning (So) of Tata Communications and Geoffrey (Holan) of TELUS, in a recent roundtable call pointed out that data center networking and the data center operator play an increasingly critical role today in providing end-to-end service.

Why is the Data Center Operator Critical?

When we think of how the end-user uses services today, the chain actually starts from the end-user and ends at the data center. Why? Because an end-user (e.g. a wireless user at the end of the access segment, or a broadband user at the end of the cable operator’s HFC network) sits at one end of the network, typically in what we’ll call the “first mile” access segment.

The applications that such an end-user uses, however, typically reside today in the servers and virtual machines of a data center that happens to lie at the opposite end from the end-user, in what we’ll call the “last mile” back-end access segment (see the figure below). This is because applications such as VoD, itunes, and HR or financial applications, all run in large remote data centers, because of the enterprise outsourcing its traditional in-house data centers to managed data centers.

The end-to-end picture, therefore, looks like the following: a request from an end-user sitting on a wireless or wireline access network, goes through a LAN-WAN,  wireless or wireline, middle-mile network, and finally into the last-mile data center network to get serviced. From this, it’s evident that the data-center network, its architecture, performance, and resiliency/uptime have become key to delivering service to the end-user, making it an essential part of the end-to-end picture.

Simplified View of the End-to-End Network and a Distribution of Operators/Providers Across the Network Segments

What Type of Data Center Operator Are We Talking About?

Specifically, we are referring to the multi-tenant, multi-service data center operator, who is a major player in the end-to-end picture today. Examples might be: Terremark, Savvis, Rackspace, or even Tata.

How Does One Characterize the Multi-Service, Multi-Tenant Data Center?

A multi-tenant, multi-service data center  is one where the compute and application-hosting resources serve as the host platform to a number of different users (enterprises, for example, or, in the case of a data center dedicated for internal use by the operator, the different departments of the operator), and a number of different applications.

This means that any customer can come and virtually subscribe to any of the data center’s servers, and load their own applications on it. Thus, that customer’s enterprise can access those servers and applications from anywhere in the world. So monitoring the service and measuring performance now involves the data center network, which becomes a critical part of the infrastructure delivering service to the end-user.

Within the data center, the operator has a choice of technologies TRILL or SPB (801. 11aq) or VX-LAN or other proprietary technologies, such as from Cisco and other vendors. The data center network itself comprises multiple layers – gateway switches, Layer 2 and Layer 3 switches, aggregators, End-of-Row, Top-of-Rack switches, physical servers and virtual machines. The granularity at which service may need to be monitored is all the way to the application or session level, or the level of the individual user. Regardless, the data center is part of the end-to-end network.

Why is such a Data Center Operator a Key Operator Type?

There are 2 reasons for this:

  1.  Today, the data center operator plays an essential role in end-to-end service, as explained above. This is because of the migration of enterprise applications from what used to be a single, hosted enterprise data center, dedicated to that enterprise (and, in some cases, even to specific applications) to a multi-tenant, multi-service data center, which hosts a wide variety of applications from a number of different enterprises.
  2.  The performance of different applications and different users in the data center has become critical. Again, because of the migration of important enterprise applications into the data center network, the performance of the data center network affects end-to-end performance. E.g. A delay in serving up payroll data or financial data impacts an enterprise’s business performance, and is, in many cases, directly correlated with the network’s end-to-end performance.

Examples Illustrating How this Data Center Type is Critical

Here are two examples to illustrate this point:

  1. The first is when data centers are used internally by operators, as is the case at TELUS for example. In TELUS’s case they have migrated their SIP servers and VoIP equipment to generic hardware in their data centers. So, there is no dedicated voice equipment anymore. As a result, the performance of their voice service depends on latency and end-to-end jitter, which now extends all the way into the data center. This is clearly a case where the SLA’s that the operator signs with end customers are dependent on data center performance.
  2. The second is when a large operator, e.g. TATA Communications, offers a multi-service, multi-tenant data center. In this case, multiple enterprise applications from many different enterprises are hosted there. As a result, the end-to-end performance of those applications, which now depends on the performance of the data center network, is essential for the end-user to experience good performance.

This explains why the data center and the data center operator have become critical to any expression of end-to-end performance management.

A related thought from Aamer Akhter of Cisco was that not only are the data centers becoming multi-tenant and getting virtualized, but that such virtualization extends right down to the application and the hardware

With application virtualization, we have applications that can support multiple customers, because they are running on different virtual machines. Likewise, developments such as Hypervisor, are leading to the virtualization of hardware that is raising interesting questions about performance management and monitoring, extending right to the application layer. This is an area we did not yet delve into, but presents very interesting challenges for performance management.

What is your view about the importance of the data center? About the state of application virtualization and how it impacts our thinking about performance management and monitoring? Do share your thoughts below or by writing to us at perfmon@metanoia-inc.com, and we’ll be glad to consider your questions/inputs for the upcoming roundtables and panel. We look forward to having a dialog with you!

 To read the next step of this evolution — namely, issues in considering the network’s impact on application performance, click here, and to look at the previous step, namely, the rare behind-the-scenes look, click here

To see the related vblog post, or to download an audio podcast on this subject, click here.

Telecom professionals worldwide may attend the panel and participate in all of the on-going dialog leading up to the panel and beyond, simply by registering at http://bit.ly/LknqIs.

If you or your company would like to get involved with this initiative, and contribute to it, please contact Dr. Vishal Sharma at vsharma AT metanoia-inc DOT com or call +1 650-641-0082.

The companies cooperating in this initiative are: 

Co-operating Companies                                   

————————————————————————————————————————————————–

Metanoia, Inc. has consistently been a leader in bringing the eco-system’s focus on carrier-centric issues. If you would like to contribute to, participate in, or have a suggestion about our recent initiatives, write us at initiatives@metanoia-inc.com or comment on this blog. To be involved in the current panel, write to Dr. Vishal Sharma at vsharma@metanoia-inc.com or call +1 650-641-0082.

Our industry-leading Provider Network Health Assessment Service is an amalgamation of a decade+ of experience working in the carrier-ecosystem, and uniquely designed to deliver strategic and technical expertise to operators via a series of flexible Service Packages. For details, and a representative case study visit http://www.metanoia-inc.com#NHAS. To reserve a Strategy Session to brainstorm your needs, reach us at experts@metanoia-inc.com or +1 650-641-0082