Virtual machine sprawl is a well-known challenge with established best practices to prevent and resolve. But there’s a new kind of sprawl to worry about: network sprawl.
Network sprawl is a phenomenon resulting from the growing adoption of the cloud, and subsequently, hybrid IT, where some applications and infrastructure remain on-premises, but some are transitioned to the cloud. According to SolarWinds IT Trends Report 2016, nearly all (92%) of the IT professionals surveyed said adopting cloud technologies is important to their organizations’ long-term business success, yet 60% say it is unlikely that all of their infrastructure will ever be migrated to the cloud. Thus, hybrid IT is the reality for the foreseeable future.
As this happens, the number of networks our organizations rely on -- those we own that are on-premises combined with those owned by cloud and SaaS providers -- is growing rapidly, and potentially unchecked, as more and more infrastructure moves to the cloud and business units implement more SaaS applications.
At the end of the day, however, we network engineers are still responsible for ensuring the performance of all the network connections our organizations rely on, whether we own the networks or not. In essence, we have become responsible for our networks, as well as the networks of cloud and SaaS providers, and the networks of our ISP and the ISPs they rely on as well.
This probably makes you nostalgic for the good old days, when we didn’t have to look much beyond our walls for anything except WAN links and the internet. Sure, some of us had to deal with multiple data centers and cross-site data and applications, which required dragging data across valuable wire bandwidth, but we could either ask for an application instance to be moved or mirrored if things got too bad.
But today, “data dragging” -- the purposeful design of systems so that large quantities of data must move across the network as they're processed -- is happening more frequently. And we’re not even dragging data across reliable and secure, albeit much slower, links such as a leased line or well-paid WAN links from site to site. Although many SLAs point to “T1-class bandwidth services, data is brought across the pure internet without the MPLS link to Azure or T1 to Amazon.
Effects of network sprawl
In IT, there is a smaller scale and differentiated version of Maslow’s Hierarchy of Needs. Its three important elements to success, in this order, are: responsibility, accountability and authority. Without a doubt, every network administrator assumes the responsibility and accountability for a network’s performance, but authority is less of a given, especially in the era of hybrid IT. Without authority, it's a challenge to ensure network maintenance and performance.
Indeed, the worst possible effect of network sprawl is having cloud-based applications dependent on multiple networks over which we have no visibility into and thus no authority over. These applications may range from simple things such as a website or remote web service, all the way up to complex, mission-critical cloud-based applications.
Network sprawl also has an effect on on-premises environments. Many organizations choose to keep critical IT components on-site rather than trusting them to the cloud outside the boundaries of the workplace; the core database is a good example of a fundamental IT function that’s nearly always kept on-premises. But then developers build microservices or even full-blown applications that reside in the cloud, but also rely on data from the core database. So, the inability to see or manage the performance and availability of the provider’s network, or understand why there might be downtime, will naturally affect these types of implementations, even though the users and the data are all local.
Moreover, as we move further in the world of cloud, SaaS and hybrid IT, tried and true tools such as traceroute are becoming obsolete, as they are typically blocked from accessing service-provider networks. At the same time, cloud monitoring tools don’t have visibility into on-premises infrastructure performance -- nor should they, as some would argue. So, the environment is simultaneously becoming more opaque and complex. In other words, while the network sprawls, the tools to manage the network sprawl at the same pace. A typical day at the office requires multiple screens filled with dozens of application windows, and an increasing number of nights are tied up with change- control windows as we spend hours patching and maintaining the tools themselves.
Containing network sprawl
There are a couple things you can do to manage the impact of network sprawl. The first step is admitting you may have a network sprawl problem. Find out what cloud vendors and SaaS applications your organization uses, the performance requirements needed from these vendors, and if they’re living up to those needs. Recognize that as the network engineer, you ultimately have responsibility to ensure not only the networks you own, but those of the cloud and SaaS vendors your organization relies on are performing well.
WAN providers, cloud vendors and SaaS application vendors will never hand over authority in the sense of giving you control over their networks for a variety of reasons. However, visibility is almost as good as authority. Having visibility into their networks from the outside also provides you with the authority (i.e., credibility rather than control) you need to ensure your organization isn’t negatively impacted by a provider’s poor network performance. New tools can help provide this visibility and simplify troubleshooting.
Hybrid IT is the new reality. Network sprawl is a new challenge. Admitting you may have a network sprawl issue, or least the potential for network sprawl is the first step to recovery. Gaining back the authority lost to cloud providers through visibility into their networks will give you control over your complete environment -- all the networks your organization relies on.
Leon Adato is a Head Geek at SolarWinds, an IT management software provider based in Austin, Texas. Adato boasts more than 25 years of IT experience, including 14 years working with systems management, monitoring, and automation solutions for servers, networks, and the web. Patrick Hubbard is director of technical product marketing at SolarWinds, and also a Head Geek. Hubbard, who joined SolarWinds in 2007, has more than 20 years of experience in product management and strategy, technical evangelism, sales engineering, and software development, at both Fortune 500 companies and startups spanning the high-tech, transportation, financial services, and telecom industries.