APMdigest asked industry experts to outline the most important factors that impact website response time. The third installment of the list, featuring factors 11–15, covers the back end ...
In a recent survey of 300 application developers, conducted by Boundary, we found that nearly 60 percent of participants had been affected by a Cloud outage. Around 72 percent of participants experienced significant costs from Cloud performance issues: thousands of dollars per incident and/or in excess of $100 per minute of downtime. This isn't stopping companies from moving to the Cloud, of course. The same survey found that 67% of developers say that their company is hosting “business-impacting” applications in the public Cloud.
Other surveys show similar concern around performance in the Cloud. The Cisco 2012 global Cloud computing survey indicated that Cloud application performance was one of the top three challenges for companies in migrating applications to the Cloud, after availability/reliability and device security.
It's easy to point the finger at the hosting companies. They're managing the infrastructure, so ultimately they must be responsible for performance, right? Not so fast.
Running services on the Internet is not foolproof. Whether due to weather, natural disasters, equipment failure and/or operator error, outages will occur. Large IaaS vendors, such as Google, Rackspace and Savvis, are operating highly interdependent, complex services based on dozens of data centers, broadband connections and thousands of servers around the world; 100 percent uptime is simply not possible. Plus, third-party providers can't see into your environment; they don't know what contingencies are playing out on your own network, third-party APIs and services that are being used or in the code that you wrote.
It's up to companies to fill in the gaps where their hosting partners will inevitably fail. And doing so requires a different type of monitoring capability than in years past. An industry luminary Michael Biddick, recently wrote about the need for a new generation of APM tools which can effectively monitor all components of the application and supporting infrastructure, including system and network performance. Next-generation APM systems must locate the underlying component causing the problem, he writes. Finally, APM systems working alone or with complementary products must suggest or take corrective action to resolve performance issues before they affect users.
This is sound advice. It's rare that one solution can accomplish all of these tasks. Most companies, including many of our customers, rely on multiple monitoring tools which work together and share information for quick identification of issues and resolution.
Importantly, these tools must be able to bring visibility across Cloud and hybrid Cloud environments. This dynamic, virtual infrastructure has proven difficult or even impossible for older legacy APM systems, designed for physical infrastructure, to manage.
As a result, systems, application and network groups often point fingers at one another, and waste time, while still not identifying which component is causing the issue.
If your company is using a legacy APM product and has invested a lot of money and time into it, you may be loathe to replace it. That's a valid consideration. It's worth talking to your vendor to determine how they can support your move to the Cloud. Will an update be coming soon to address Cloud monitoring? If not, can their product easily work with newer tools, to bridge the gap? But in general, new architectures demand new solutions.
Another trend is that APM tools are now offered as a service, just as the applications they monitor. This reduces the burden on IT to support yet another piece of software or appliance, and enables organizations to get up and running quickly on new monitoring systems as needed.
We are seeing a huge resurgence and growth in the APM market - causing a number of analysts to publish in-depth studies around market segmentation and needs. Companies want to monitor their IT infrastructures from an “application first” or top-down perspective, which is rendering traditional bottom-up tools as legacy. Something that everyone appears to agree on is that application monitoring is not a one-size-fits-all situation and customers should understand their requirements fully before selecting their partners. The good news is that with tools being offered as SaaS and on shorter-term subscription contracts, the cost of adoption and change has lowered dramatically.
We are seeing modern applications and Cloud computing drive huge growth in the new generation of solutions while traditional/legacy solutions are withering away. We are also seeing a clear distinction emerging between developer-focused solutions and operations-focused solutions, as follows:
Developer-focused solutions answer the question: “Where in the code is my problem area?” If the problem is not in the code, then of course these tools offer limited help.
Operations-focused solutions answer the question: “Where is my problem?” These tools must cover 100% of your environment but don’t go as deep in code analysis.
It’s a transitional time for the APM technology market. More than perhaps ever before, companies are realizing that to succeed in the massive change of placing IT services in the Cloud, an investment in comprehensive and always-on monitoring tools is a must. Otherwise, the Cloud can backfire. Users and managers will not quickly forget if their apps crash or sensitive data is lost forever. Selecting a next-gen APM tool today that is designed for monitoring modern, distributed, Web apps and services will help a company best prepare for a transition to the new enterprise computing environment underway right now.
ABOUT Gary Read
Gary Read, CEO and President of Boundary, previously served as CEO of Nimsoft, providers of the award-winning Cloud monitoring solution, where he grew the business from zero to over $100 million in bookings and 300 people. As CEO, Gary led all aspects of the company including product, marketing, sales, support, and finance, guiding Nimsoft to a successful acquisition by CA for $350 million. Nimsoft experienced significant worldwide growth, with approximately 1,000 customers in 36 countries. Prior to Nimsoft, Gary held executive positions at BMC Software, Riversoft, and Boole and Babbage.