Managing cloud resources together with on-premises data centers is making life hard for IT teams.
Usually, hype about an industry term comes with at least some benefit for IT. To varying degrees, big data, SDN, IoT, BYOD and even hyperconvergence are being pushed by vendors who can demonstrate at least some ROI or leaps in technology that enable new business opportunity. That’s not the case with hybrid IT. True, it’s a fresh buzzword de jour, and vendors are selling components of hybrid IT, but they aren’t selling hybrid IT as a technology. That’s because hybrid IT is not something we asked for --it’s the side effect of vendors selling cloud and SaaS. So, it’s alright to seriously dislike it.
It’s OK to call the baby ugly
For transparency purposes, I work for SolarWinds, and we make IT software. But I also manage a hybrid IT infrastructure, and in speaking with hundreds of customers in the same situation, mine is fairly typical. About 70% of our critical business web applications are SaaS, and 20% of our internal apps have migrated to Amazon and Azure. Cloud supports not just machine images, but acts as an endpoint for micro services, CDN and elastic capacity. Rolled together, however, that’s maybe 25% of our total infrastructure, and like most of you, the other 75% is in my data center, spindles whirring as it has for eons.
Before SaaS, with the exception of email and a little web browsing for the marketing team, we only worried about WAN to support branch offices, and could rely on SNMP to manage just about everything. As we expanded support from networks to applications, we added new skills and certifications and our monitoring and management systems grew along with us. But the basic approach -- enable, root, sa access based polling -- was the same, because we had visibility into everything that might make the phone ring at 2 a.m. Complex troubleshooting may take time on-premises, but there are no dark corners.
With hybrid services, we now find critical services at a minimum behind a veil and more often behind a wall -- with barbed wire and a killer rabbit. We were told we were buying hassle-free services that would simplify our lives as admins, but too often we’re now working end-user experience helpdesk tickets with newly opaque service components we can’t dissect.
The other area of frustration with hybrid IT is a significant learning curve. If you’re a born-in-the-cloud IT organization, you’re comfortable with open source, elastic resources, containers, service redundancy, durability and more. You’re also a DevOps-first shop, and manage much of your infrastructure programmatically. And to be sure, this is a Good Thing that exorcises many IT entrenched evils.
But for those with hybrid infrastructure, you’re still primarily managing established and increasingly complex enterprise technologies. Because the majority of your operations are in the data center, you’re more likely still interrupt and break-fix driven, upgrade via months-long capex processes, and have learned more application and hardware specific technology than you cared to. In this case, your introduction to hybrid IT came weeks or months after you began implementing cloud and SaaS components.
No network or application engineer worth his or her salt would ever admit being overwhelmed with technology, but most privately admit that there are only so many new technologies that one should have to juggle in a day. Admins don’t want to wield a dozen vendor-specific dashboards, they don’t want to receive alerts from a dozen different monitoring systems and they really don’t want to cobble together reports for management on spreadsheets because that’s the only method for metrics aggregation and planning. Yet many admins are as they remember the sales guy saying it wouldn’t be a problem.
Sometime soon, we’ll solve our compliance and risk roadblocks with cloud and significantly depopulate our data centers, akin to IBMs old “Who Stole the Servers?” commercials. It will be the second wave of abstraction after virtualization, the first taking us off metal and onto hypervisors and this time moving those workloads to commodity infrastructure for efficiency. We’ll be DevOps engineers with a couple of gray beards around to manage the campus WiFi and gripe at ISPs when the WAN gets jittery. Yes, the future is always so bright and shiny.
In the meantime, however, it’s OK to be skeptical when cloud and SaaS vendors promise painless integration with our current infrastructure. For now at least, we have servers to run, home-brewed apps to tend and mailbox datastores and databases to migrate. Our heads will expand to co-inhabit both realms and deliver coherent services even if the bits are bifurcated on the back end. We’ll accept interim hybrid IT, just as long as it’s not forever.