From the blog.

Managing Digital Racket
The more I tune out, the less I miss it. But that has presented me with some complex choices for a nuanced approach to curb
Complexity – My Friend, My Enemy
Over my years of network engineering, I've learned that the fewer features you can implement while still achieving a business goal, the better. Why? Fewer

Network Engineering In A World Of Data Center Haves & Have Nots

1,663 Words. Plan about 11 minute(s) to read this.

In December 2014, IDC published a piece entitled “IDC Reveals Datacenter Predictions for 2015.” Quoting the piece directly, I’ve selected four trends as particularly interesting from a network engineering perspective. Let’s look at each and consider the implications on the assumption that these predictions are correct. Which…they aren’t likely to be. Predictions never are. Even so, there’s some worthwhile thought provocation going on here.

Trend 1.

In the next two years, incompatible or immature IT asset management practices will prevent 80% of organizations from being able to take full advantage of converged and software defined IT solutions in their own facilities.

Or stated bluntly, IT operations & accounting practices won’t change quickly enough to be able to leverage modern computing techniques. The issue here lies in IDC’s words “converged” and “software defined.” These imply a reorganization of IT, where silos are abolished, reorganized, or somehow tightly coupled so that all IT technology disciplines work together.

There’s also a trend towards faster replacement cycles for gear. Rotating equipment out in 2 or 3 years instead of 7-10 seems plausible when the capex is low enough and the technology is changing so rapidly. Shortening the replacement cycle is a problem if the accounting depreciation process isn’t keeping up.

The implication for network engineers.

If IDC is correct, most organizations aren’t keeping up because technology is changing too quickly. To maintain competitive advantage, some businesses will choose to house their compute in a way that gives them the ability to do whatever their customers are looking for quickly. That might mean running their applications on some cloud provider’s infrastructure.

Shifts will be industry specific as well as case-by-case. However, if you’re in an IT group that is neither evolving processes nor working closely across silos, you’ll be buried with the rest of the fossils. Does that seem like hyperbole? I’m increasingly of the opinion that it is not, for one simple reason: money. Public cloud will get to the point where it will make sense for businesses to host there, instead of hosting their own applications. When the ROI is finally there, keeping IT in-house will seem like an outmoded way to think unless IT operations makes the changes required to make full use of emerging technology.

Of course, this doesn’t mean businesses won’t need networks. More on this later.

Trend 2.

By 2016, hyperscale datacenters will house more than 50% of raw compute capacity and 70% of raw storage capacity worldwide, becoming the primary consumers/adopters of new compute and storage technologies.

Interesting thought. Where will the most compelling places to work be, assuming you’re a person who loves bleeding edge technology? At places running “hyperscale datacenters,” which means just really, really large DCs using scale out techniques to handle massive data loads. This doesn’t mean just Microsoft, Google, Amazon, etc. I believe this also includes very large business operations like eBay or financial service companies like JPMC. Any sort of organization that needs to both crunch a lot of data and serve a lot of customers has a business reason to run at hyperscale.

The implication for network engineers.

Nothing. Or everything. This comes down to the individual. What sort of a shop do you want to work in? Some shops will have unique requirements for their applications that force them to in-source their compute infrastructure. Some shops adapt slowly. Some, like certain government organizations, might be cash-strapped in such a way that they just can’t change. You might be okay with that.

On the other hand, you might want to be in the middle of the latest and greatest technology, perhaps helping to pioneer some new techniques yourself. In that case, perhaps you want to find some hyperscale shops and seek your fortune. Clearly, that’s where much of the action is for practitioners.

Trend 3.

Over the next two years, over 60% of companies will stop managing most of their IT infrastructure, relying on advanced automation and qualified service partners to boost efficiency and directly tie datacenter spend to business value.

There’s a key word here that would be easy to overlook considering the other points. That word is “managing.” IDC’s not saying that the infrastructure goes away necessarily. They are saying that businesses will outsource management of that infrastructure to people and processes that can do an efficient job of it.

There’s a lot to be said here, because my experience is that outsourcing IT infrastructure is often a better idea on paper than in reality. Businesses might think outsourcing is a good idea because they have an adversarial relationship with their IT department, or think of IT simply as a cost center to be reduced and optimized from an accounting perspective.

A good IT department head will act as a liaison between the business and technologists, helping the business to understand why their IT team is valuable, and conversely help their IT team understand how technology is used to accomplish the goals of the business. When the business views IT as a partner facilitating business rather than an ever-growing expense, outsourcing options will be scrutinized much more closely. Without active communications between IT and the business, the business will be more likely to have a shallow, “black box” understanding of their technology infrastructure, oversimplifying it to line items on a budget. In addition, if IT staff seems incapable, unhelpful, or otherwise incompetent, the business won’t trust them to run the infrastructure well.

The implication for network engineers.

If infrastructure management is outsourced as predicted, that means it might make more sense to work for the firms taking over the management, as opposed to the end consumer of the technology. Anecdotally, I know more and more resellers are interested in managing infrastructure for their customers, as this is another revenue generating opportunity. I get e-mail from various people in this space who are exploring new products they can offer to their customers. This makes sense in networking especially, as this area is exploding with new products and technologies that existing businesses might find difficult to use effectively, especially if the IT department finds itself at odds with the rest of the business.

Trend 4.

Over the next three years, 70% of large and mid-sized organizations will initiate major network redesigns to better align inter-datacenter and datacenter-to-edge data flows.

Networkers ask me fairly often what technology they should focus on next to stay relevant and marketable. This IDC prediction is a compressed answer to that concern. A few points to ponder.

  • Who’s doing the major network redesigns? Large and medium businesses.
  • Why are they doing a redesign (not to be confused with a simple upgrade)? To make sure that the DC is moving data around in ways that makes sense. “Inter-DC data flows” comprise traffic between data centers, an issue for those business with multiple DCs that move their compute around, often in the context of active/active DCs. “DC-to-edge flows” are those flows leaving the data center and going to remote clients. Is the network optimized for this traffic? There are many considerations here that bleeding edge networking techniques are chasing down (think NFV and service chaining for starters).

The implication for network engineers.

To remain relevant and marketable as IT changes, study the problems that larger organizations with data centers have, learn how those problems are being addressed, and see if you can get a job working on one of the network redesign projects. Easier said than done, to be sure. Still, having an awareness of industry issues and comprehension of how those problems are being addressed puts you ahead of the curve in terms of vision. Skills such as network device configuration and security policy management are more valuable when you understand how they fit into the context of business needs. Big picture thinking will go a long way to getting you through an interview related to network architecture.

The coming great data center divide.

The IDC predictions point towards a consolidated world of compute. If we assume economies of scale, eventually, it may become silly for a business to own lots of IT infrastructure. Why not lease it from cloud providers? They’ll be able to do it cheaper, and besides…they’re experts. I think it’s possible that businesses will eventually migrate most (if not all) of their applications to the cloud. Now, we can argue that point. There are several plausible reasons that this might not happen. But if this hypothetical mass migration comes to pass over time, what does that mean for networks in general?

One point to keep in mind is that networking never goes away. Ever. Public cloud is useless if it is not accessible. Therefore, campus and enterprise networks will continue to thrive. In addition, the WAN will also be key. Many businesses run their WANs to allow remote offices to access centralized data resources. If the centralized data changes locale, the problem of access remains. So, the WAN might look a little different, but there will still be a WAN. Remote access from individual clients might look a little different, too. But the requirement of fast, reliable connectivity will remain.

Enterprise and campus networking doesn’t go away as a discipline just because data and compute loads moves. That said, there is a huge question of whether businesses will need to retain one or more full time network engineers. It could be that network engineering talent will move out of enterprises and into VARs that specialize in building and maintaining access layer network infrastructure, a point that was alluded to in the IDC predictions.

If IDC is right, there is a great divide coming between the hyperscale DCs (the haves) and everyone else’s compute infrastructure (the have nots), but there will still be lots of technology work to be done. As always with technology, practitioners must constantly evolve to keep up.

Or go to the beach. Hey, you can always bring a book to study…

palm-trees-ethan-banks