As we enter more deeply into 2014, we should look more closely at the compelling argument for extending applications into the cloud/hybrid cloud to address the need for fast, flexible capacity. According to comScore (News - Alert), $42.8 billion was spent online from desktop devices; an increase of 10 percent compared to a year ago – a clear example of Web-based systems temporarily needing extra capacity, while having the same needs for security and control.
Hybrid cloud is a powerful architectural concept. Properly managed, it can help IT organizations smooth capacity requirements, even in the face of large fluctuations in demand. But while it can be relatively easy to implement, hybrid cloud-based distributed applications aren’t so easy to manage. New visibility issues and increasingly complex applications create the potential for hybrid cloud applications to actually perform worse than those that only run in the data center.
The challenge of ensuring application performance is making operations teams reluctant to evaluate new hybrid cloud environments, and subsequently, slowing deployments. Firms are interested in gaining the benefits, but are hesitant to move forward. Everyone is nervous about introducing this great new technology into their operational environment because of one word – visibility.
Performance and availability management live on visibility. Proper visibility allows application problems to be found and fixed quickly. Operating without that visibility is the equivalent of driving without a dashboard (or a clear windshield). Eventually, you’re going to run into something, and it will grind everything to a halt.
End-to-end Transaction Visibility
So what is the proper visibility for distributed applications? Taking a service-oriented approach, the IT operations team responsible for maintaining application performance has to focus on the delivery of business services:
- Are they running?
- Are they responding in a timely manner?
- Are they acting correctly?
Answering these questions for business services requires end-to-end transaction monitoring, with hop-by-hop visibility across every server/machine within the application infrastructure. Only then can IT operations understand where transactions go, where they get stuck, and why. Complete transaction visibility also allows operations teams to solve many application problems without calling on technology silo experts to dive into code.
In hybrid cloud environments, the need for end-to-end transaction visibility actually increases, even as operations teams intuitively think that they have fewer systems to worry about. Ultimately, they must still deliver proper business service performance, which now run on systems both in the data center and in the public cloud.
It’s a New World
For IT operations, the key to performance management in hybrid cloud environments is being able to tie actions in the public cloud-hosted systems with the connected actions in the data center. With a host of “Cloud Management” tools to pick from, this one function is the most critical for ensuring performance and availability.
Step 1: Map and monitor transactions in the Data Center
The first step is to make sure you have a view of transactions in your own data center (or “on-premises”) systems. The dynamic nature of distributed applications means that tools should be able to adjust their transaction maps automatically – updating as individual application components are added and dropped from the overall system. (Most experienced IT operations professionals will tell you that manual mapping efforts will be out of date before the Visio diagram finishes printing.)
Step 2: Map and monitor transactions in the Cloud
Remember, IT operations is still responsible for business service delivery. Just because a server resides off-premises, the responsibility doesn’t change. Visibility of cloud-based systems is absolutely necessary for operations teams to maintain control.
Step 3: Tie them together
This is actually the most difficult part of the transaction management equation. There are some tools that can track transactions in the data center and some tools that can track transactions in the cloud, but for IT operations to understand the end-to-end performance of a complete business service, they must be able to stitch a “login” transaction that started in a public cloud-based Web server with its specific “login” counterpart over in the data center.
With an end-to-end transaction view, IT operations have the ability to see everywhere transactions go, know whenever they get stuck and isolate the location of any problem (whether in the data center or the cloud) quickly and accurately. Only with this capability can IT operations support the hybrid cloud distributed application environments that will continue to grow and change.
Edited by Maurice Nagle