Confio Software has for years successfully sold database performance management solutions to DBA’s of Orace, SQL Server, DB2 and Sybase databases. Confio’s Ignite product line uses an agentless architecture to query a comprehensive set of metrics from these databases, allowing DB’s excellent visibility into the true performance of the database – and especially visibility into latency based issues.
Earlier this year, Confio added IgniteVM to its product line. IgniteVM queries the vCenter API’s (nothing new here – lots of products do this), but then combines vCenter’s view of storage I/O latency, with a view for the DBA of database I/O latency.
Why is this important? As we all know, a very high percentage of the performance problems that occur in virtualized environments have to do with I/O. It is also obvious that the vast majority of the I/O in the environment is created by database servers like Oracle, SQL Server, DB2 and Sybase. Finally it is also the case that many organizations are reluctant to virtualize database servers precisely because of these issues. Frankly it is the fear of unknown performance problems that is at the root of this reluctance – a fear that is fed by a lack of information that allows for database server I/O latency to be benchmarked before virtualization, and then allowing for comparison to those benchmarks to be made after virtualization.
IgniteVM neatly solves this problem by giving the DBA a complete view of the latency and resource utilization metrics at the database layer, the virtual machine layer, the physical host layer, and storage layer. If you think about IgniteVM as an Applications Performance Management (APM) solution focused upon the database server as the application, IgniteVM breaks new ground in the APM industry by combining infrastructure latency information (which is the crucial way in which performance has to be measured for virtual infrastructures) with application layer (the database) query latency and response time information. IgniteVM also tracks changes that occur on the vSphere host where the database server is running so if a new VM shows up that creates a problem, you get an event in IgniteVM along with the spike in latency. A screen shot of this is shown below.
We have stated before that in order for performance critical applications to get virtualized, that the team that owns the virtualization environment will have to step up and assure the performance of these applications. This is easier said than done. There are few tools that measure applications response time for every application across tiers irrespective of how they are architected (BlueStripe and AppFirst are two such tools). However since just about every performance critical application is based upon a SQL database, picking the database layer as the point of management might be a good choice as well. Confio IgniteVM is uniquely positioned to provide such an approach to both DBA’s and owners of virtual infrastructure.