The Cloud: Ready for Enterprise?

The “cloud” seems to now be the universal term for just about any off-premise solution whether that is PaaS, IaaS, SaaS, etc. There has been much debate about whether the concept of cloud is truly enterprise ready or not, with most of the conversation being started by cloud providers or those that would directly benefit from cloud use. Traditionally they have tended to focus on 4 main “promotion” points of the cloud, its cost, scalability/performance, security and uptime. Lets look at these points further:

COST

A large virtue of the cloud are the amazing cost reductions that can be had. Truth be told this is NOT a “one size fits all” assessment and will greatly depend on the business, its needs and objectives.

Many seem to believe IT just magically vanishes and is no longer needed in any capacity any more which is completely false. If your running your department correctly, your business should see your group as an essential driver of tools/processes to enhance the business and not just the server/computer people in the corner. Part of the allure of the cloud is a fast, agile environment that is able to respond to the needs of the business, I guess the cloud does make sense if your group is not currently meeting this need for the business. A 2013 IT group should be though, as we have the knowledge, tools and ability to move projects forward for the betterment of the business. The cloud doesn’t change any of this, taking the raw resources of the cloud and making them work for the business whether in your cloud or their’s is still an IT function. The only time you remove staffing expense is to outsource which is a different discussion; your title may change and may change what your everyday “glass” looks like but a business still needs a person or group to plan and deploy solutions for the business. Personally I don’t think staffing costs play much into the cloud cost discussion, you may change some positions or be able to move the SAN guy to another project but you will likely not get rid of them. Look at NetFlix as a prime example, completely cloud but they still have a group of DevOps folks who look after everything.

I won’t go into an exact inhouse vs cloud pricing table as none of that matters; my environment isn’t your environment, my contract negotiating skills are different than yours so we won’t pay the same thing even if we had the same thing. What you really need to do is understand what makes sense to move to the cloud (if anything) and calculate those costs. This will require you to actually identify your inhouse expenses, which i’m sure you already know right? If you don’t you really should, makes these and other discussions easier with the finance folks if you can talk actual numbers. Don’t just move to the cloud cause its cool and “everyone’s doing it”, only even consider it if you have a need: do you currently have a need to scale quickly for a project for a short term or is your workload predictable? Generally speaking the cost of the cloud doesn’t make sense for predictable, stable workloads. Ideally you have subscribed to the virtualization and standardization best practices and already can offer your business a flexible, cost effective, reliable infrastructure inhouse, if you can’t the cloud may be a good fit. What you will find is most cloud providers will be extremely expensive compared to your inhouse infrastructure, especially when it comes to storage and bandwidth costs. A big challenge to the cloud providers is the fact commodity hardware prices continue to plummet; case in point, price out 10-24TB of local storage versus the cloud, if your good at negotiating you could have that storage paid off in 6-12 months of cloud prices by the time all the extras are factored in. Also many folks seem to forget that bandwidth transfer is “free” at home but not in the cloud, plan on paying for that! I think these numbers become more in favor of inhouse as your business grows, many early adopters are now discovering this. Zynga, HubSpot, Eli Lilly just to name a few have left cloud services recently, eHarmony has recently left AWS in favor of inhouse collocation. The cost of the cloud usually isn’t in your favor over the long haul; same concept as leasing or buying a car really.

Now I am not completely against the cloud, I can speak pretty well to the costs because we have priced out clouds options for many of our recent projects and in “most” not all cases inhouse was the better route, some examples:

– We examined Exchange onsite, Office365, Rackspace hosted email, Zimbra hosted and Zimbra onsite when we were considering our move from Exchange 2003. We chose Zimbra onsite for a variety of reasons, one of the highlights was cost though. Our Zimbra deployment paid for itself in under 6 months.

– SPAM filtering. This is actually one of the easiest “no-brainer” cloud choices we’ve made. We looked at onsite solutions and it made no sense for the equipment, cost and complexity it would add to our environment when you stacked it against a hosted solution. This is one case where we clearly chose the cloud, have saved lots of money, maintained 99.999+ reliability and benefited from the providers global spam database.

– Before upgrading our SANs we did examine IaaS solutions. We looked at AT&T Synaptic, iLand and AWS. Without getting into specifics, our internal infrastructure costs are low, our growth stable and predictable and being 100% VMWare virtualized made it so that a cloud provider could not touch our internal infrastructure cost so we kept everything inhouse. With the cost of replicating our current environment in iLand we could have purchased new hardware every 6-12 months and still come in cheaper!

– When we were examining project management solutions we looked at in-house Bugzilla, Pivotal Tracker (cloud product) and solutions like MS Project. In the end we went with Pivotal Tracker because their product had better features, was fairly priced and fit well with our development culture.

SECURITY

This is one place the cloud has NOT proven itself at all. Each cloud provider runs its own setup and policies about data, there are really no best practices yet fully established which is why in my mind this deserves an F grade. Most cloud providers will not talk in depth about any of their policies/procedures for their services. There is no true auditing abilities from the providers end: how do you know a copy of your VM instance wasn’t made, can you prove it?

I also really don’t pay much attention to the concept “they have more resources and thus my data is safer”. I do agree in some situations there can be value added to services by size, think large SPAM filter companies who have a huge customer base to write signatures from, but I don’t think the cloud has necessarily proven this. I also think due to the cloud’s age it really hasn’t had to fully prove itself yet, I think once more and more large profile businesses begin getting data into the cloud this will provide more incentive for the “walls to be tested” by those who would have something to gain.

I like to think of the cloud like flying commercially; the actions of others aboard the plane can have direct impact to you even if you are doing everything right. To me this fits perfectly with the cloud, maybe your servers are secure and your best practices are the best, how do you know that your “neighbors” in the cloud are doing their best and not endangering your shared environment that you live on? I think this is  a larger issue with multi-tenant environments not just cloud specifically.

UPTIME

This is another area where I definitely think the cloud has not proven itself a bit. When you look over the past couple years of the main players in the space they both have had serious events contributing to hours upon hours of downtime. Most concerning is that many of these issues have human or automation process failures; Microsoft services had issues due to expired security certificates, Amazon couldn’t fail over as it should have, etc. In my opinion these are bugs of an untested system in its infancy, not something I would entrust my 24/7 data needs with. Google/VMWare of course is too new to the market to truly have any uptime/operational data to speak of. Most of these services and the companies that rely on them have not been recently achieving at least 3 9s over the course of a year. If NetFlix goes down at Christmas its kind of a bummer cause you couldn’t watch TV but when your company goes down for that amount of time is a different story. A silly service credit does NOTHING to reimburse the credibility loss or the actual financial loss that can easily be calculated when a business is unable to work for periods of time. My other concern is the great lack of support during these events; you literally are customer #234543 in the cloud waiting for the shared resources to come online. There also tends to be a SEVERE lack of root cause analysis and permanent resolution publicly addressed when these issues happen, although I do have to give major points to Amazon, they will usually do a blog post outlining these items but that is definitely not the norm.

Scalability/performance

This is perhaps the clouds biggest/best “positive” argument. If you truly have the need to go from 100 servers to 10,000 tomorrow morning for a big launch there is no place like the cloud that can deliver that capacity that quickly. The problem is the high premium you pay for the instant access to those kinds of resources. Depending on the size/needs of your business this may or may not be an issue.

Related to the cloud’s performance, I think this is yet another area that has not been proven. Unless you have the backend access to the storage, networking and servers you really are not getting a “clear” picture of the performance nor do you have the ability to generate logs/data to defend yourself in situations in which you aren’t getting what you pay for. Alan has a chilling blog post about this issue here: http://alan.blog-city.com/has_amazon_ec2_become_over_subscribed.htm

Obviously some of these issues you have in a virtualized environment whether its in-house or in the cloud but the real difference is in-house you have the access to the backend systems to view important details, can collect logs and address any issues. In the cloud though, that level of access really doesn’t exist nor do you have a true picture of the load your particular host is experiencing. If you have issues, minus the folks at your cloud provider, who can you turn to, and what if they say everything’s fine?

Outlook

I think the next few years will be interesting to watch, the cloud market is going through HUGE changes, most important the “race to the bottom” in terms of cost, which will kill many of the smaller, more niche competitors over time which I think is Google, Microsoft, VMWare and Amazon’s long term goal, grab as much market share as you can and then once competition is dead slowly increase the costs again once those customers are “stuck” in your cloud environment. For the more “traditional” IT environments that would embrace the cloud I think VMWare offers a very compelling situation as the toolsets/abilities of your in-house technologies are then extended to the cloud, Microsoft will eventually offer similar value to Hyper-V folks. Google/Amazon will continue to appeal to the “web 2.0+ beyond folks” who make use of their services today.

I think those that think the cloud will be dead in 2 years are naive, it is most definitely a disruptive opportunity for those that can justify the costs/trade-offs, does it spell the end of IT, No, is it a “NEW” technology, No. Its basically the combining of a whole host of technologies and concepts together in one big marketing train. The technologies that power the cloud are still your standard network/storage/compute pieces and the concepts of scale being applied on a large scale. The cloud really reminds me of shared web hosting back in the late 90s early 2000s, once someone decided you could share those resources of a server it took off and the providers “raced to the bottom” to offer cheap hosting, end result is now you have $4.99 a month web hosting with all the fixings, does everyone do shared hosting, No, do some still run in-house services, Yes, did shared hosting change the industry, Yes. Cloud computing will be no different, there will be those that jump aboard the train, those that never get on and those that get on and then get off and go back to in-house. The group that will lose the most are the smaller consultants, vendors and “resellers” who will be unable to differentiate themselves from their parent vendors who they are now directly competing against. Great example: VMWare cloud resellers (iLand) will REALLY have an uphill battle as now VMWare is offering the same services as they are, why go with a 3rd party when you can subscribe to the company that CREATED the virtualization market and get support from one large vendor who also probably handles your in-house virtualization needs? Some thankfully have recognized they will never contribute anything substantially to the market, read Dell and have just embraced a “diet” cloud option living off their partners and providing some integrations between them.

WAIT….Travis you hate cloud providers and thus your article is bent towards in-house services. NOPE, your incorrect. I have actually done my research and examined all options for our projects including hosted/cloud options, presented those findings to my management and let them decide. In MOST, NOT ALL, cases our in-house services provide better cost, uptime, features and flexibility than the cloud can. BUT, there have been clear times where we chose the cloud because it made sense from a money, services or uptime standpoint (see some examples above). I really feel some IT folks really need to embrace the “best outcome for the company” mentality, sometimes it may mean the cloud/hosted solution is the best, other times in-house is clearly the winner. Creating an environment that meets the needs of your company should be your primary concern and quit focusing on the tools to get there, the cloud, your in-house servers are all tools that should serve your business, if the business begins serving your tools then you have it wrong. I think the cloud has gained a lot of traction in the non-IT space because it addresses issues that have been prevalent in IT for years; inflexible, not meeting needs of the business, excessive cost, etc. The cloud should not be something to be afraid of but should be viewed as another tool in IT’s toolbox to meet the needs of the business. Knowing your business, its needs AND the costs of doing business all help in determining whether the cloud is ready for YOUR enterprise!

Advertisements

Author: Travis Kensil

Director of IT. Husband and father. Michigan beachbum.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s