Solve performance degradations with real-time analysis and metrics

Performance degradations are a fact of IT life

Performance degradations have a significant impact on the productivity of your internal clients using the business applications.

There may be times when application performance degradations last not just for minutes, but for hours, days or even weeks.

With today’s technology, you can give your IT team the ability to solve downtime and degradations in minutes and with much lower budgets than in the past.

How can we do this? With real-time analysis!

Let’s look at network recorder’s history

For many years, the “dinosaurs” of network performance analysis have relied on good old-fashioned technologies (now obsolete – read more in this article): it consisted in capturing and storing gigantic amounts of packet data. These “stream-to-disk” or “network recorder” solutions will basically swallow and store terabytes of packet data. It is, then, down to you to find the time window that is necessary to analyze the data. In a nutshell, it’s a needle in a haystack.

To diagnose intermittent performance degradations that have been going on for several days or weeks, they claim that tens of terabytes of traffic need to be stored; that you have to filter the data you load for detailed analysis, which, obviously, requires you to know precisely when and where the problem occurred.

This will lead them to sell you expensive appliances, while you are in fact mostly buying disk space, rather than network intelligence.

What real piece of intelligence are you being sold?

The concept of these solutions is to just store, but then the analysis becomes tedious, time-consuming and sometimes simply does not provide an answer.

The work required to analyze such volumes of data is prohibitive unless you have a very powerful analytical engine to process the data. Since the data gets stored as it goes along – as raw data, no previous selection and computing have been undertaken. Even if you have accurate information on the approximate timeframe of the problem, you’ll still spend a lot of time looking for the needle, if you can even find it!

In terms of intelligence, you’re only being sold a search engine and disk writing speed, so where is the intelligence in it?

Today the real alternative is real-time analysis!

Rather than storing packets and then attempting to analyze them when needed, real-time analysis consists of automating packet analysis and computing analytics as traffic flows through each of your networks.

As a matter of fact, you will:

• Save most of the manual analysis by getting manageable and understandable performance analytics, available any time within seconds.
• Extend your visibility to a much wider network scope
• Store your analytics instead of raw packets and save most of the memory space needed.
• Accelerate your analysis by sorting, filtering, viewing an already computed and comprehensive set of usage and performance metrics. 

Intelligence = Know How + Automation + Scalability

What intelligence are you being sold: an analysis « know-how » of different protocols, from the lower layers -IP, TCP, UDP- up to the application transactions HTTP/HTTPS, SQL, SMB, DNS, Microsoft services, Citrix… at very high throughput, higher even than the storage of raw packets on the hard drive.

Why haven’ these solutions replaced these dinosaurs? Unlike a solution that preserves packets, these solutions required years of development, and it has taken time for them to stand out because they were created by start-ups. It isn’t disk space that you’re being sold anymore, but “know-how” or insight, engineering, research, and development.

The result is a simple solution that can be used by non-specialists and enables faster troubleshooting.

The advantages to your team include:

• Data retention
• Much more precise baselines
• Incomparable granularity: all information is available to the precise minute for over 1 month
• Access to troubleshooting in just a matter of seconds
• Lower acquisition and possession cost
• Easy to use tool for your teams

And if you’d like to get further information about this topic, I’d like to suggest you read our White Paper on real-time analysis