Data Centers: The next five years

Last week at the annual Gartner Data Center Conference in Las Vegas, key analysts overviewed the most important issues that data centers face in the next five years. As expected, rising energy costs and the movement toward “green” solutions was a primary topic, as was social networking technology. But there were several other hot topics which the responsible network operations professional should be pursuing.

Below is our take on eight of the top 10 issues, as raised by Gartner. As a refresher, the top 10 issues were virtualization; the data deluge; energy and green IT; complex resource tracking; consumerization of IT and social software; unified communications; mobile and wireless; system density; mashups and portals; and cloud computing.

1. Virtualization – As virtualization technologies continue to advance, adoption is accelerating. Savvy data centers have already begun implementing virtualized servers, and this growing trend is here to stay for a variety of reasons including dynamic server load management, a need to reduce power consumption, and fast growing network complexity.
2. The data deluge – Not a new issue, albeit an accelerating one, the vast amount of data that traverses today’s networks creates a variety of problems, from storage management to throughput to monitoring. And here’s the kicker…the problem will only continue to expand over time. Faster and more efficient network infrastructures (10G today, 40G and later 100G in the future) will help with throughput. Gartner’s automated tiering idea for storage holds water strategically, but not all data centers are poised to implement it in the near term. Monitoring is particularly concerning, as higher bandwidth tools are required, yet virtualized servers and applications can be a challenge to pinpoint for monitoring in dynamic environments. Advanced capabilities for data/traffic aggregation and filtering are not only “nice to have” in virtualized environments, but a must.
3. Energy and green IT – Not much to say that you haven’t already heard. When it comes right down to brass tacks, this is a cost saving move, and fortunately (for your corporate communications division), it can be leveraged for some positive PR. Given current global issues related to energy costs and environmental concerns, this trend is here to stay.
4. Consumerization of IT and social software – This is an issue that has grown significantly over the past year or two, and one that Gartner has said will continue this accelerated growth path over the next several years. Consumers are very interested in new technologies, and especially in the ability to control their own user experience. Coupled with a growing trend toward a distributed workforce, work-from-home privileges, and the vast volume of unstructured data that result from social web-based applications, we see consumerization as inevitable (and in some respects, a good thing).
5. Unified communications - UC has been touted as the grand vision for bringing all communications media together, and it most certainly has its benefits, idealistically. That said, the growing trend toward digitization of communications is a prerequisite to truly achieving this vision. Kudos to Cisco and other thought leaders in this area, because as we analyzed in our post about the Net Generation, real-time and cross-platform (even cross media) integration is exactly the type of capabilities the younger sect of the population desires.
6. 

   Mobile devices like the Android now function like mini-computers

   Mobile and wireless – Continuing on the idea that the younger generation wants pervasive connectivity and interaction, mobile and wireless are growing in importance by the day. This is yet another way that social technologies are forcing our hand at considering new strategies for accommodating user needs. Coupled with the desire for “real time everything”, we have seen mobile phones advance to the point where they can serve as mini computers. The key issues then are how to secure devices that are not physically attached to the network itself, how to accelerate performance over time as mobile network speed increases, and how to properly provision devices and connections.  Brace yourselves, because the complexities that this will pose may not even all be known as of today.

7. Mashups and portals – Both similar in nature, in that they serve important content aggregation purposes, mashups and portals offer different challenges due to the nature of how they work. Of particular concern to us is the mashup concept, where not only content but code can be aggregated. The obvious problem is that you risk exposing yourself to untrusted code, not to mention potential copyright issues related to syndicating content that is not authorized for distribution and/or republication. Since browsers were not designed with mashups in mind, this is an issue that absolutely cannot be overlooked from a security standpoint.
8. Cloud computing – With all of the cost and efficiency benefits that cloud computing offers, cloud security remains a wild card. Two of the most common arguments about cloud security center around the access / authorization process and storage / transmission of sensitive data. Access can be a problem for anything that falls outside of your physical environment, but that is a lesser concern in our eyes than securing the actual traffic stream that transmits across the ultimate cloud…the Internet. It will be fun to watch how this all plays out in the coming months and years.

So there you have it, our take on eight of the 10 issues that Gartner said will be the most important to data centers heading into the second decade of this century.

We’d love to hear your thoughts on the issues and our commentary. Feel free to agree or disagree. Debates can be very useful in helping hone our viewpoints on all of these issues, so I truly hope this post sparked a new idea or thought for you.

The hidden cost of network monitoring
Troubleshooting is a major culprit in the productivity equation. Network operations staff members frequently find themselves in “firedrill” mode, chasing the emergency of the day. When a problem arises, someone must physically make and break connections in the data center to connect and disconnect tools. This is a very time-intensive process that must take place before troubleshooting and again after remediation. According to various industry estimates, this form of troubleshooting can take between 25% and 60% of a typical workweek for data center operations staff.

Filtering is also a major issue. Because the network changes daily, data center staff would need to be continuously adjusting filters to ensure complete coverage and optimal tool utilization. Clearly, this cannot happen with tools alone, particularly when troubleshooting is an ongoing priority.

A second filtering issue relates to overlapping packets, i.e. packets that meet the requirements of multiple filters. Most filtering approaches operate sequentially, where one filter runs and the results are sent along to a tool, then a second filter runs on the remaining data, and so forth. Any data that should go to multiple tools does not, because it was already sent to the first tool. This can be manually adjusted, but it typically requires the operator to identify all overlaps, sort out what data should go to which tool, and write complex filter rules in a complex Command Line Interface (CLI). CLI coding is a major headache – simple overlaps can typically require 4-5 lines of code, and the numbers increase exponentially as overlaps increase. In most cases, this step is simply overlooked, and monitoring coverage is compromised…meaning that 80% visibility could actually be closer to 50% in a real deployment.

Without budget to add headcount or fund advanced cross-training for staff members, data centers are forced to make tradeoffs in priorities, and more often than not, the firedrills win.

Monitoring Optimization is the best solution that solves all three of these problem areas.

First, optimization solutions are fully integrated with GUI front ends, making them the easiest-to-use monitoring tool optimization solution on the market. Everything is managed via drag & drop within the GUI, providing significant efficiencies for setup, management, and especially troubleshooting (simply re-configure connections in the GUI, with no make & break required). Not only do these GUIs offer intuitive interfaces, but they are also where all filters are set, without requiring any complex filter rule coding or other CLI-based management. The GUI provides a full top-down view of all ports, tools, and filters.

Products like the Anue 5200 Series Net Tool Optimizer also offer Dynamic Filtering, the first of its kind. Dynamic Filtering automatically accounts for overlapping filters upon initial filter definition, ensuring all of the right data makes it to the tools that require that data to do their jobs. This filtering approach is dynamic in its ability to self-maintain for changes to the network, without requiring manual intervention on a repetitive basis. Dynamic Filtering alone can save days of work for set up, filter maintenance, and ongoing oversight.

Next: ROI for Optimizing Staff Productivity (Case Study)

Optimize your monitoring to easily and effectively get the most out of tools

By now, you’ve surely seen the Anue Systems announcement of the Net Tool Optimizer generation of products. Aligned with that announcement, you’ll also notice that the term Tool Aggregation is no longer representative of everything the products can do, hence the new moniker Monitoring Optimization. Let’s spend a few minutes reviewing exactly what we mean by this technique – and that’s exactly what it is, a technique for optimizing your network, application, and security monitoring approach, methodology, and execution.

At its core, Monitoring Optimization is about achieving much more than visibility across the network segments and applications. There are a range of classes of products that can provide some or all of the visibility you desire, including everything from simple point solutions to some of the more sophisticated switching, TAP aggregation, and data access network products. And with the right expert managing those systems, visibility is only a matter of time and effort. But that’s only part of the story…

Tell me something – if several classes of products can provide roughly the same level of solution to the visibility problem, how do you decide what to buy? There are many details and features that can be compared, such as type of filtering capabilities, connectivity options, various bandwidths of operation, widely varying form factors, etc. In reality, this whole discussion should be a moot point, because when it comes right down to it, it is a battle of “Speeds and Feeds” as we once referred to it in the semiconductor industry. As soon as you build the faster buggy, someone will come along and leapfrog you (or even themselves, if that were somehow within the realm of physical possibility), by building the next “greatest thing since sliced bread” contraption.

In the real world, you need to take a step back from the noise and look at what is really important. Visibility is now a tablestake for even being in the game. How can you as a network operations or security professional make sense of the whole thing?

Simple – look at the entire monitoring equation. Monitoring Optimization expands beyond visibility to include two more variables:

• Utilization of the tools that monitor the network
• The productivity of the staff responsible for managing this whole operation

This is where many solutions fail to deliver. Everyone is stretched financially these days. With the hefty investments you make in tools to monitor your environment, you absolutely, positively must derive the maximum value out of those tools. To do so, the tool utilization has to be carefully managed to prevent overload, which can lead to dropped packets and compromised visibility/coverage. On the flip side, it makes no sense to purchase expensive equipment, only to use a fraction of its capabilities when there is ample monitoring capacity available for the taking.

Ease of Use is Critical To Monitoring Optimization

To top this all off, your chosen solution should be very easy to set up and manage. With shrinking support staffs and an ongoing demand for “more from less”, is it really feasible to adopt a solution that requires many weekly hours of effort to manage static, CLI-based filter coding methodologies? Or even worse, who can afford to bring on a full-time CLI-coding expert to manage the system?

A lot of vendors are introducing GUIs these days, but you need to be critical of what that GUI delivers you. For it to truly be easy to use, the filtering must all be engrained in the GUI, and not through some trickery where you can reuse CLI-based code within a pretty package. Remember, someone has to write that code in the first place, and we’ve seen reports where CLI-based filtering can add up to four hours of setup time depending on the sophistication level of the user. That’s four hours of your life which you will never get back.

Let’s leave it at that for now. We’ll go into more detail about this topic in future blog postings. Please do not hesitate to comment below or reach out to us directly to learn more or debate these topics. We’dd appreciate your feedback and discussion.