Part of making the most of Application Performance Management (APM) is to recognize its limits. BYOA and outsourcing make for a few of those boundary cases that define the best use of APM.
“BYOA” is the proliferating initialism for “bring your own application“. Information technology (IT) “consumerization” involves not just employees walking in with their tablets from home, and expecting to connect to the corporate LAN (local area network); now the point is that they’re using Yahoo mail or Google Docs instead of Exchange and Microsoft Office.
What’s the significance of BYOA for APM? Not so much, I think. In principle, APM is plenty flexible to add, say, LinkedIn messaging to its dashboard as a monitored application, even though most of the implementation is outside the using organization’s control. Certainly modern APM needs the technical ability to track end-user experience (EUE) that crosses corporate firewalls more than once and integrates services across those boundaries.
Should APM aim for this universality, though? No, I think–not because of technical limits, particularly, but because of business strategy. Ultimately, we don’t deploy APM from an aesthetic compulsion to track everything happening on our networks, but for the specific benefits APM brings. It’s just not to IT’s advantage to track Instagram performance, as long as Instagram is an unsupported BYOA-class application. APM, and the expertise to run it effectively, are better invested in the portfolio of services and applications which IT already covers.
There’s a more positive way to say this, as consultant Joe McKendrick hints at the end of the article referenced at the top: IT needs to incorporate BYOA into its “governance” policies, and be ready to make BYOA one part of a successful, comprehensive support plan. That almost certainly means that IT concentrates more on in-house applications hammered by 400 workers simultaneously at the expense of one convenience used by a single vice-president–but it also means readiness to enhance support of applications which first appeared as BYOA but eventually proved useful to a large portion of our service population.
Jonathan Feldman’s recent extended plea to “[e]mbrace software-as-a-service” (SaaS) reinforces this perspective that there are limits to APM. The Real User Monitoring blog most often focuses on APM’s potential and techniques. At the same time, it’s healthy to remember that the ultimate goal is to help our end-users achieve the most possible, and that IT is just a means to that end. Applications with little unique value, or those with little variance in controllable performance, might not be worth monitoring. Just as it’s time for us to accept growing use of SaaS and BYOA in all the places where those alternatives are good enough, it’s time for us to ensure we focus APM first on the applications and services that make a difference. To improve performance of a key application by 38% through clever APM-based diagnosis is a real achievement. It’s also a great accomplishment to solve a workflow or provide a service entirely outside IT. Feldman identifies that the point is to get “… the right mix … [of] core competencies … [with e]verything else … a candidate for selective sourcing.”