Application Performance Management (APM), is defined by Wikipedia as “the monitoring and management of performance and availability of software applications. APM strives to detect and diagnose complex application performance problems to maintain an expected level of service.”
When approached correctly, APM makes both the data center managers and software developers look like rock stars. However, when Application Performance Management is neglected, or simply an afterthought, that’s when the wheels begin to fall off.
And that’s the reason why the Data Center World show managers at Penton and AFCOM likely decided to schedule a panel session on APM: to promote a healthy understanding of the importance of this discipline to data center professionals. Or as the session description hinted at, to combat the blame-game that pings-pongs back and forth between IT -- responsible for hardware and configuration -- and developers -- responsible for the code.
Data Center World Panel on the IT Operations Management Track (ITOM)
I attended the Application Performance Management (PANEL SESSION), which was part of the IT Operations Management (ITOM 1.1) track at Data Center World, held at the Ernest N. Morial Convention Center in New Orleans.
The panel was moderated by Jason Shepard, Managing Principal of Cresa MCS.
(I’ve followed Shepard on social media for quite some time and thought of him as one of the country’s top go-to experts in data center real estate. I had no idea he was so well-versed on application infrastructure that runs in data centers.)
The panelists included:
- Bernd Harzog, CEO of OpsDataStore
- Dave Wagner, CTO of OpsDataStore
- John Parker, Global Data Center Manager of ESRI
21 Key Takeaways from the Panel on Application Performance Management
- There is no way any single vendor can do it (properly manage applications and infrastructure without APM). [Wagner]
- The role of the CIO is changing to that of a cloud-wrangler, managing disparate systems, responding to the pace of innovation in the industry. [Shepard]
- Virtualization has had a massive impact on the data center industry and APM. Consider when 20,000 square feet of data center spaces goes to 20 racks and then into a colocation facility. [Shepard]
- There is ever-increasing pressure to improve the quality and business performance of online services. Compete online or die. Competitors own no data centers and no infrastructure. Insurance companies, banks, and retailers are competing with virtual versions. The pressure to evolve software is unrelenting and ever-increasing. There is pressure from end users and business units within enterprises. This pressure results in a tremendous amount of innovation. [Harzog]
- Docker was invented to get software into production more quickly. To move from development to test to production without touching code. The problem continues to be more challenging. [Harzog]
- When problems happen with AWS (Amazon Web Services), they’re less than useless. [Harzog]
- Virtualization is increasing density, appealing to the cost-conscious, and now a factor with AWS. [Shepard]
- AWS is like a late night infomercial. By the time you get done paying for all of the AWS add-ons, it gets very expensive. When you spend $10 million to $20 million annually on AWS, you hope to be down to $6 million annually in AWS spend next year. [Parker]
- With AWS, there are no guarantees of application throughput and response time. And there is no way of measuring and controlling this in a public cloud. [Wagner]
- The economics of public cloud are like the economics of crack cocaine. It feels really good on the first hit but feels really bad later on. The economics of the public cloud are in every way superior to colocation or your data center. But public cloud doesn’t give you any information about what’s going on with their hardware. [Harzog]
- If they (public cloud providers like AWS) catch you writing code to try to instrument and measure their hardware, they’ll disallow you as a client. [Wagner]
- Measure what actually matters, as opposed to measuring silo by silo. Measuring the apps themselves is the only way that matters. [Wagner]
- The IT side of the house looks at this differently than the infrastructure side of the house. [Shepard]
- Application Performance Management is hugely siloed. You need an ecosystem that spans those traditional divides (IT and infrastructure). [Wagner]
- The job of an APM tool is to tell you what’s wrong with the code. But, what if the problem is not in your code? Try calling up AWS and tell them there’s something wrong with the cloud. Many are completely missing large layers of information about what’s going on. [Harzog]
- SaaS and edge are the future of software, pushing applications closer to the user. [Shepard]
- When looking at IaaS vs. SaaS, make sure that applications are portable back from public cloud to colocation. Many have had to impose some discipline on developers (to make this happen). [Harzog]
- There is a lot of good instrumentation out there, but no way to glue it together. [Wagner]
- Partner with the right people in the right part. By creating partnerships, it will make it possible to build analytics to solve problems; by having learned from earlier failures. [Harzog]
- “What is the time to value?” is a tremendous challenge for management tools vendors. How much time and money is it going to take to get value? If you focus on this from the beginning, it eliminates 80% of Application Performance Management tools very quickly. You have to be able to get value in less than one year. [Harzog]
- Because Gartner requires instrumentation within the JVM (Java Virtual Machine), if there are apps that you bought, classic APM as defined by Gartner is useless. [Harzog]
Were you at this Data Center World panel discussion? Or are you a big fan -- or big critic -- of Application Performance Management? Please share your thoughts in the Comments below.
If you're a fan of APM, you may also want to check out AppOptics from SolarWinds -- which helps developers, ops engineers, DevOps consultants, and IT pros stay on top of every level of the full stack.
And if you're in the data center, mission critical or cloud services industries, or you sell to the data center industry, don't miss our weekly update newsletter -- Data Center Sales & Marketing Institute (DCSMI) Update Newsletter. Get notified about new reports, events, podcasts, and blog posts.