Monitoring comes in many, many forms today: application, networking, infrastructure, data center, performance, virtual and now cloud. These terms pop up, often without distinction or acknowledgement that this new type of monitoring is not really new at all, but is rather a rehash of a much older “flavor” of monitoring. The explosion of terms to describe monitoring has more to do with the number of monitoring vendors, and more to the point, those vendors’ marketing departments, than it does with new forms of monitoring emerging ...
In APMdigest's exclusive interview, Matthew Ellis, IBM Vice President of Service Availability and Performance, discusses APM including cost concerns, APM in the cloud, and Gartner's 5 dimensions of APM.
APM: What have been IBM's most important advancement in APM in the last year?
ME: One of IBM’s most significant innovations was the introduction in 2011 of an agentless transaction tracking solution that works in harmony with our existing agent-based solution. This combination, which is unique in the market, gives our customers the best of both worlds – the ease-of-use and time-to-value of agentless tracking combined with the detailed information provided by an agent-based solution in the domains that need it. Agentless and agent-based data combine seamlessly to provide our customers with incremental value and a complete picture of their application transaction topologies.
APM: What is the secret to successful APM in the cloud?
ME: There are three keys to insuring application performance in cloud-based infrastructures:
- Visibility beyond the firewall
- Robust SLA monitoring of public and private cloud infrastructure
- Tight integration to traditional monitoring
Getting performance data on individual cloud components is crucial to rapid problem isolation and diagnosis, but is often hindered by incompatible (or non-existent) instrumentation or an inability to share data in a meaningful way. Effective SLA monitoring involves watching every transaction that crosses the firewall boundary, and alerting when expectations aren’t being met.
Lastly, since moving applications to the cloud is a process, and very few IT divisions are 100% cloud-based at this point, it is critical that the APM data you get from the cloud tightly integrates with your existing traditional management solutions.
Ideally, you want an APM solution that is completely infrastructure-agnostic – you have exactly the same visibility, presented in the same way, whether the application is running natively on physical hardware, on an internal virtualized infrastructure, in the cloud, or some hybrid combination of all three.
APM: A recent study from Trac Research shows cost management as a key APM concern. How does an organization find the right balance between how much money and time they can afford to spend on managing applications versus how much visibility they can get?
ME: For each organization, the investment in APM is going to vary. Of course, it is ultimately an ROI discussion. For some, any incremental amount of increased visibility increases confidence in their support of critical applications and can be justified in improved availability or optimized performance of critical applications. For others, there is a clear point of diminishing returns where further investment is no longer warranted.
We recommend a staged approach to APM deployment that allows simple, high value goals to be achieved rapidly and enables further investment in greater visibility to be seamless and incrementally added.
APM: What are the steps you recommend?
Many organizations start by simply monitoring the application response times that customers experience to ensure that application behavior is meeting their expectations.
The next stage is to deploy our agentless transaction tracking solution which can monitor applications across the infrastructure without investing in deep metric evaluation of all involved application components. The information learned with this part of our APM solution can show where applications are spending most of their time, and suggest where richer instrumentation would be most beneficial.
At this point we recommend installing local agents for deep monitoring of critical components to collect all of the information that can be important to maintaining optimal application behavior. Some customers opt to install deep monitoring on all of the components of their critical applications, and some go even deeper, capturing information sufficient to enable application debugging of production applications.
Different organizations and different applications have different needs. By providing a multi-layered APM solution that progresses through very simple steps from response time monitoring to different levels of transaction tracking and even application diagnostics, IBM is able to provide a solution that can be easily deployed and extended incremental for the most demanding organization.
APM: Why do you feel the Gartner Magic Quadrant on APM named IBM as a Leader?
ME: IBM has a comprehensive vision of APM. IBM’s APM solution offers a combination of proven technology, industry-leading integration, and extensive breadth of coverage. In addition, IBM’s continued focus on ease of use, rapid time-to-value, and role-based pricing and packaging make our portfolio straightforward to adopt in production environments.
Gartner defines APM as having 5 dimensions: End-user Experience Monitoring, Discovery, Transaction Profiling, Deep-Dive Component Monitoring, and Performance Analytics. A unified solution incorporating each of these dimensions is critical to insuring application performance, by enabling the context for action that is so critical to modern operations.