Nmedia - Fotolia
Cloud computing offers a plethora of benefits and enterprises are deploying cloud applications in mounting numbers. Spending on cloud infrastructure and services should hit $174.2 billion by the end of this year, according to IHS Technology Inc., a 20% increase from 2013. And as the cloud footprint grows, data centers face the daunting task of managing these systems.
But the management software needed to provide clear visibility into cloud application performance is immature, forcing businesses to make a trade-off. Extend existing infrastructure to the cloud or manage cloud-based systems with less visibility than they are accustomed to with other apps.
Traditional management tactics vs. cloud management complexities
Enterprise management tools typically operate at the nuts and bolts layer of the infrastructure and examine items such as operating system interactions and database management system transactions. Most technologies run on top of this layer and tools concentrate on the big picture -- overall system performance, typically response time or system latency. However, the different design points create several challenges for data center pros.
Enterprises gather integrated information from a number of point management tools, allowing them to present one consolidated view of system performance. These tools examine specific elements, such as server throughput, data moved by storage I/0 or network bandwidth. Instead of overwhelming an admin with measurements from multiple places within a large enterprise network, businesses roll up performance information into central management tools such as IBM Tivoli and Microsoft System Center.
These tools deliver high level snapshots of infrastructure performance and notify IT teams when problems occur or, ideally, before they become an issue. Admins can then drill down into the problem and take necessary steps to rectify it.
Limit cloud-based management complexity
Cloud adds complexity to management. Enterprise IT needs to integrate another layer of performance data into the central system. And while this may sound simple, this step is challenging for several reasons.
Cloud systems typically sit a layer or two above the database management system and OS levels, as well as run on top of the virtualization layer, so they do not offer the easy drilldown information that IT operations personnel rely on to troubleshoot performance glitches. To close the gap, IT teams often build a link from existing management tools to the cloud management tool. And this step can be complex. Cloud systems include various management APIs, but they sometimes lack the necessary functionality.
Cloud management tools also do not follow typical data center equipment nomenclature. Many enterprise troubleshooting tools depend on host names and IP addresses to identify devices. However, cloud tools often rely on virtualized system naming conventions. If these differences aren't translated, the admin may have limited visibility into the infrastructure.
Some companies connect to a public cloud service, which introduces other variables. First, it requires one additional network connection -- from the company's data center to the cloud. As a result, the business needs to extend its management system with software agents on items, such as network routers, data center servers and storage to see what's happening on those devices. The extra link can also cause performance latency. Enterprise IT needs to know where the cloud transaction is processed -- where the vendor's data center is located -- and the effect that distance has on overall performance.
About the author:
Paul Korzeniowski is a freelance writer who specializes in cloud computing issues. Based in Sudbury, Massachusetts, Korzeniowski has been covering technology issues for more than two decades and can be reached at [email protected].
Application performance management aims for new heights
Analyzing the multitude of Azure management tools
Cloud sprawl a killer to cloud performance strategies