Don’t know about function services? It’s time you did.
Connecting state and local government leaders
Function services, which provide computing to develop, run and manage applications on demand, can be ideal for both inward and outward facing cyclical government services.
Even as government agencies strive to modernize their information technology, the state of the art keeps moving. In the last wave, “virtualization” eliminated the need for each application to have its own server, leading executives and legislators to believe that they could consolidate data centers full of underutilized servers. Now, a new “function-as-a-service” product class has emerged to enable serverless computing. It is important for government CIOs to understand how this new service fits into their modernization plans, especially in light of new Data Center Optimization Initiative (DCOI) guidance.
Serverless computing with function services are now offered by the largest commercial cloud providers under various names, such as Amazon Web Services Lambda, Google Cloud Functions and Microsoft Azure Function Services. Function services provide computing to develop, run and manage applications on demand. Without function services, applications run in the cloud with the same complexity as on-premise virtual machines, incurring labor and compute charges similar to an agency data center.
But what happens to data center administration complexity and costs? Using function services, the application comes on-line when needed, goes off-line when not needed and automatically scales to meet demand -- saving both computing and administrator costs.
Function services can be ideal for cyclical government services, both inward and outward facing. Examples of public-facing and cyclical service are tax administration, licensing and grant reporting; inward-focused services include payroll and year-end financial consolidation.
To back up for a moment, the motivations for modernization remain constant. Administering stovepiped legacy systems grows ever more expensive. Constituents want and deserve the levels of online, digital services they’ve come to expect from the best of the private sector. These two factors push agencies get leave the IT infrastructure model and free up resources for modernization that contribute to the mission.
The tech advances and customer demand have given rise to two highly intertwined federal policy streams that have just been updated: DCOI and the cloud smart (formerly cloud first) strategy.
The revised DCOI acknowledges several conditions and establishes some new mandates for federal IT that align with function services. Moreover, in revising its IT performance metrics, the Office of Management and Budget has recognized that data center utilization is no longer a useful metric. This raises the question of how far the government can go in taking advantage of function services for serverless computing.
At the same time, and more importantly, policy-makers revised the server utilization metric for reporting compliance with the Federal IT Acquisition Reform Act. OMB acknowledges the cumbersome effort and associated costs of managing virtual servers in government’s real-world operating environment. Critically, the DCOI memo recognizes that ensuring availability without threatening overload-induced crashes “is best achieved through agile system management practices.” Besides DevOps and DevSecOps, these practices include “dynamically and rapidly assigning resources based on measured usage and projected need.” Here the memo specifically describes the exact types of cyclical applications that are best managed using function services.
OMB ties the revised DCOI directly to the cloud smart strategy, calling on agencies “to identify applications that require large amounts of resources seasonally or infrequently which may be ideal candidates to shut down or move to cloud technologies.”
Plus, OMB adds the metric of availability, uptime and planned or accidental downtime, for federal data centers. The policy doesn’t explicitly say so, but what better way to get away from poor availability reports than by taking advantage of serverless computing?
So how can IT managers use function services to meet data center consolidation and cloud smart metrics? They should keep in mind that with function services, as with other classes of commercial cloud products, different providers are suited to various types of workloads. Moreover, containerization is needed, and different applications will be able to take advantage of different function services. That’s why agencies should consider working with an engineering-oriented industry contractor and require that partner to specify the cost-performance tradeoff for different cloud services providers’ offerings so they do not miss out on benefits by simply moving their old stuff into a new house.