chaos monkey
enterprise cloud

AWS Needs To Figure Out Its Enterprise Plan

Next Story

Publishing Startup Graphicly Raising $1M More As It Aims For Profitability

RodneyRogersEditor’s note: Rodney Rogers is chairman and CEO of Virtustream. He is also on the boards of UnitedLex, Teliris, and Greensmith Energy Mgmt., and is an angel investor in Cloud Sherpas, BetterCloud, iYogi and others. Follow him on Twitter @rjrogers87.

I’m a large enterprise and my ears are ringing. I hear that you, AWS, know exactly what I need.

Before I get into that, AWS, I want to let you know that I admire you. You made a market. You, quite brilliantly, made and own the public commodity cloud IaaS market. It’ll be awhile before Google GCE makes a dent, and most of the carriers and technology OEMs scrambling to build their OpenStack commercial extensions are far behind you. You own the public commodity cloud, and you will continue to do so for the foreseeable future.

But this doesn’t mean you know me, AWS, and I believe your foray into enterprise will expose you. Yes, I realize this may be cloud blasphemy to some. Here are six pro-tips from yours truly, me, your Fortune 500 prospect:

1. First Off, re:Invent This

At the most basic level, your enterprise messaging is naïve. There was a lot of bravado coming out of Las Vegas. Here’s what I heard: Private cloud is bad; hybrid cloud shouldn’t exist. The only answer is AWS public cloud, and you’re a fossil if you don’t realize that. You’re talking to me like I’m one of your hacker groupies.

You’ve built your business largely by selling self-service IaaS to those technologists building holistically modern, scalable, cloud-aware applications that drive new-generation consumer technology, and to support test/development environments in the enterprise. Actually, you’ve mostly sold to the 20-something kids of my IT leaders with minimal human interaction against virtually no competition. Well played – err brilliant! And… that won’t happen here. It’s not that simple. There are myriad use-case requirements within my IT walls. Please see my last post on this very topic.

2. Everyone Gets Their Ass Kicked Sometimes For No Good Reason

You’ve used your superior scale and product portfolio to dominate the IaaS space for the commodity public cloud use-case against both proprietary up-starts and OpenStack wannabes. In my large enterprise arena, however, you’ll be competing with companies that are 100x larger than you and who have enjoyed deep selling relationships into the enterprise for 10x as long as you’ve been around.

It’s not to say that those legacy technology providers are better than you. In fact, this space has never been so ripe for disruption. However, you can bet all your marbles that large legacy OEM technology (good or bad), legitimate up-and-comer innovation, irrational VAR FUD, legacy golfing relationships, et al. will make the competitive process a lot more complicated than you’re used to. You’ll need to respect that and acclimate to it. Thinking it’ll be different for you is like thinking a newly elected U.S. president will go to Washington, D.C., and change Washington, D.C. We’ve all seen how that’s worked out.

3. Get Ready To Grind, Baby, Grind

IRR, ROI, ROE, ROA. There’s an endless financial rationalization to the sales process in large enterprise. It’s built this way on purpose. Why? To slow it down, of course! This is not something you’re used to. Beyond financial justification, you’ll have to benchmark your value proposition against your competition. Get ready to get frustrated. Get ready to watch pipeline entries slip from quarter to quarter. What will seem unnatural today will be SOP tomorrow.

You also announced at re:Invent that you have hired more than 100 salesmen “to focus on enterprise.” Bravo! The fact that you thought that was newsworthy makes my point (see Nos. 1 and 2). The enterprise does not necessarily just buy IaaS if you plan to capture high-value, full-production landscapes. We buy solutions (see Nos. 4-6). Oh and finally, it takes IaaS and application content (sometimes even client industry content) to sell those pesky solutions. A “salesman” is only one pawn in the game. Think expensive customer-facing solutions engineers and architects. And more.

4. Performance Enhancers Are Legal; It’s Called Your IP

Most of the private cloud appliance pushers out there fear-monger “performance and security” when they sell their solutions to me versus the public cloud. As cloud aficionados, we all know that fear is not always justified with proper engineering. However, I see three problems here for you:

  1. All public cloud technical architectures are not created equally.
  2. Most of my workload landscape infrastructure spend today is not on “cloud-aware” applications.
  3. Your private instance offering emasculates much of the economic benefit of your core product.

You can try to tell yourself that SLAs don’t matter and are not worth the paper they are printed on. However, for revenue-generating, production-application environments, the reality is they do matter to me. I not only need assurance that the infrastructure will be available, but assurance of response times of the production applications that run on it.

Enterprises are largely virtualized on VMware today leveraging SAN storage technologies. We all know the expense of VMware licensing and the scale-out inefficiencies of its cluster limitations. However, this high-overhead configuration is built for secure performance, and this has pre-conditioned your buyer. You’ll need to prove a modified Xen kernel running simple multi-sized VMs atop NAS, as IaaS will be enough to ensure enterprise-grade performance. (Hint: A Chaos Monkey will not be the answer for sophisticated memory-intensive, transaction-intensive, fully integrated production ERP systems.) Not having an abstraction layer to optimize consumption and/or guarantee things such as IOPS (ultimately, application response time) will make this a tough road for you. Not impossible, but certain use-case requirements are going to kill you.

5. Make Your Customer Forget These Two Words: “Availability Zones”

I’m not going to rail on your highly visible cloud outages here. I believe it’s silly the way some of your competitors do so in their brain-dead “I-don’t-live-in-a-glass-house-also” sort of way. If anyone knows that sometimes “shit happens” in technology, it’s me. However, it’s important you also understand that shit can’t happen here.

The fact is that your more sophisticated customers today can, for a price, engineer ways to reduce outage risk. In the enterprise, for revenue generating production application landscapes, there is simply no alternative. You’d do well to offer pre-engineered replication/fail-over as a core service offering. We may not be hipster-cool consumer technology startups, but when you do have a problem, we can’t stop shipping product and get on Twitter to bitch about your outage.

6. Get A Little Touch In Your Life

Your “no-touch,” managed-services model is just not going to fly. I actually would suggest the opposite. Again, we buy solutions. In fact, our environments are constantly morphing. We buy and sell divisions. We require constant performance tuning and sizing. We require asset utilization rates that do not look anything like this.

Managed service wrappers sell IaaS in complex, full-landscape production IaaS. The thing is, it’s really more about application performance than infrastructure for us. Having the content to support those applications running on your infrastructure is a big-time difference maker. Partner with an SI like an Accenture or Capgemini for that? Hmm. I can only imagine what happens to the value proposition in this case (likely bad things), but I guess “one throat to choke” would be pro-tip No. 7.

In My World, The Customer Actually Does Know Best

Unlike those sexy startups, I have a lot of time, capital, and expertise invested in my IT infrastructure. Flip a switch to holistic unsupported self-service commodity public cloud? No. I’m going to use existing assets where it makes sense, leverage new technologies (potentially yours included) for the same reason, and apply those new technologies by required use-case.

There might be something to this hybrid cloud thing after all.