Posts
Cloud Computing is a scary word in some conservative enterprises, but it doesn't have to be. Here are some proven ideas on how the conservative enterprise can dip their toe in the water today, get comfortable with the cloud and save some real cash.
First, a quick 30 second primer on cloud computing for those who aren't familiar (skip what's in italics if you're already in the know). Commoditized hardware solutions now are offered in the form of Cloud Computing Infrastructure Services. Many vendors like Amazon, Google, Microsoft and IBM are starting to build massively scalable data centers (think power plants), and are offering up their computing resources in a utility pricing or “pay by the drink” model. This means that instead of companies having to pay for an entire server and all of the business continuity, disaster recovery and administration equipment/fees that come with it, they can choose to pay by the GB of storage used per month or pay by the CPU power they consume. In other words, you can pay for infrastructure services the same way you pay for water or electricity, with all of the BC, DR, licensing and support costs baked in. This is an incredibly compelling pricing structure that takes advantage of the concept of economies of scale, and it has begun to get huge traction.
The birth of these services is relatively recent, and there still are a lot of questions around security, SLAs (99.9% uptime is the current standard) and privacy, all of which are being handled in slightly different ways by the different vendors. There’s going to be standardization (and perhaps ultimately, regulation) in these areas as the technologies mature, but regardless, there are ways to take advantage of these services in the short term, without introducing risk into your organization.
OPPORTUNITY 1) CLOUD TEST ENVIRONMENTS
Test environments are often costly. Conservative enterprises often try to mimic production configurations in their test environments. They also have multiple test environments to isolate different phases of the testing lifecycle (i.e. system, integration, uat, performance, etc.). The cloud is a great place to host test environments while handing off the administration duties to the application teams. An infrastructure organization can provide a pre-configured image/virtual machine of an application's test environment to the development staff for deployment to the cloud. The app teams then have the flexibility of spinning up as many test environments as they need, and they only pay for the up-time of the server. So if you only need the test environment available for 4 hours on a Friday, you're only paying for those 4 hours! It also allows for an easier way to tie back infrastructure costs to individual applications, to allow for more granular budgeting for infrastructure services.
OPPORTUNITY 2) GEOGRAPHICALLY DISTRIBUTED STORAGE/DR
Many conservative enterprises pay an arm and a leg for offsite storage for disaster recovery purposes. This offsite data isn't often used as a part of day-to-day processing, it exists for DR purposes. The cloud is a great alternative for offsite storage, with commoditized pricing of .10/GB/month. Several utilities allow you to store data encrypted on the cloud as well.
The cloud also provides a great alternative for sharing files geographically. Many of the large cloud infrastructure providers allow you to park your files on different continents. So if your European offices complain of slow access to data, you can have a mirrored drive or database on the cloud that they can access locally, while your "source of truth" is safe within your own data centers in the States.
OPPORTUNITY 3) TIER 2/3 OR DEPARTMENTAL APPLICATIONS
The third opportunity is leveraging the utility pricing of cloud hosting for tier 2 or 3 departmental applications. If you have non-mission critical apps with non-mission critical data, those are great candidates to migrate to the cloud. Since most major platforms are supported on the cloud today (Windows 2003, OpenSolaris, Multiple Linux Distros) many custom and packaged apps can be hosted on the cloud. This is another way for enterprises to dip their toe in the water while keeping their mission critical applications safe at home.
OPPORTUNITY 4) COMMODITIZED ENTERPRISE SaaS
Many packaged enterprise applications are moving to hosted SaaS subscription-based pricing models. Microsoft Sharepoint and Microsoft Exchange are two examples of commoditized applications (i.e. non-business differentiators) that companies can outsource. The licensing models are competitive, but arguably a Total Cost of Ownership analysis should be done, since the size of the organization can often dictate the degree of savings.
These are a few examples of how conservative enterprises can start leveraging the cloud for real cost savings today. What are your thoughts? Are these really low-risk alternatives? Are there other ways companies can use the cloud to save cash without "handing over the keys"? Does it make sense for enterprises to start dipping their toes in the water at this point? Let me know in the comments below.
-J
10 comments:
All of these would be reasonable uses for cloud computing. I think that the challenge, though, is knowing exactly WHO your provider is. By this I mean that some services are quite secure, and others are certainly less so. It would be important to find out exactly how secure the provider is, and this might not be easy to do.
In addition, security could easily be compromised (accidentally or otherwise) by employees, regardless of the level of security of the provider. I use Syncplicity to store some data, and it allows you to share your files with others. One careless employee could compromise critical data through misuse of this type of functionality.
That said, I think there is a very valid place for non-sensitive computing in the cloud. It's just important to understand that it is not the same as an internal network with its attendant security and controls.
I am currently using Amazon EC2 for my startup Myfitbrain. The experience has gone pretty well so far. It was a little more complicated than first expected. I do not think I could have done Myfitbrain without an inexpensive way to put up this site with plenty of room for robust expansion in the future.
I definitely think Clouds are a way to go for the future.
I admit I'm way behind the tech curve on cloud computing. But you're fired up about it so I s'pose I'll go learn more. :)
What about small/growing businesses? Seems like there's a tremendous opportunity there for the "pay as you go" infrastructure model for the startup or 100-400 person company that isn't yet ready to internalize that cost center.
Hey Derek, I think Jim answered the question. For the startup or SMB the cloud is a no-brainer. Enterprises have been slower to adopt since they've already made the capital investment. They're probably going to have to cut the cord at some point though, the question is, when?
I can vouch for the SMB play in the cloud. With the exception of one legacy accounting client-server app, our (Solstice's) entire organization's infrastructure is running in the cloud. This includes our CRM, Collaboration/Messaging platform, Portal, Wiki and Knowledge Management System. Not to mention whenever we have to mimic a client's environment for outsourced development, the cloud is the first place we go. We can provision a server in 9 minutes. Can't beat it.
Jay, there are a ton of SaaS/BPO applications that are viable for large enterprises to utilize. As a provider of Single Sign-on/Federation technology to over 400 customers and approx 50 of the fortune 100 companies in the US my company (Ping Identity)is knee deep in exactly what you are asking about. We have customers like Morgan Stanley utilizing Salesforce.com, Google (apps and mail), and other SaaS/BPO apps across their enterprise. 3M is using Salesforce Ideas portal to generate 1000's of new and inventive improvements. Others like GE, Cisco, Dell, and Fidelity are all leveraging SaaS/BPO applications like WebEx, Concur, travel providers (Expedia,Orbitz, Rearden Commerce), HR and Benefit providers (SuccessFactors, Taleo, WorkDay), online backup and DR, etc...
The key to enterprise adoption of any cloud app is really more about committing to a paradigm shift and less about technology.
There are things like security, audit and compliance, availability, and customization that need to be consider. These are all minor and way more easy to over come than building and maintaining custom code and software stacks.
Hope this helps.
Jason Bonds
I know that the Microsoft offerings in terms of Sharepoint and Exchange offer seamless integration between who is still local in the enterprise and who is on (in?) the cloud; making it easy to pilot and than migrate existing users. Jay, what you mentioned about provisioning a server in 9 minutes is a very compelling story. Organziations can get BACK to what they do best (selling,etc)and leave some of the IT gruntwork to the cloud. In effect, this is "outsourcing" and the compelling model for outsourcing is to farm out what can be done cheaper/better by experts so that you do what you do best (sell, etc).
Cheers...
Todd Klimek
Having a test environment in the cloud does not give you an apples to apples test environment. How do you mitigate changes between the virtualized test environment causing problems when the code is migrated to the physical production system?
I like the idea of Sharepoint in the cloud. I think that comodity technology in the cloud is cost effective.
A lot of advantages of cloud computing come from virtualization, but virtualization is possible in your own data center too.
I am currently consulting at Exelon and they aren't ready for the cloud yet, but they are ready for virtualization. A central organization is starting to provide "lab managers" and "virtual machines". Via a lab manager application, a department can spin up and spin down servers on demand -- just like in the cloud. My area would like to use the virtualization to consolidate numerous test environments that are completely under utilized.
The cloud may certainly be the right solution for many, but it seems to me that any cost analysis should compare: "internal virtualization" vs. "the cloud".
What do you think?
Some great comments left recently I'd like to respond to.
To Joseph's point, he is talking about the concept of the internal or "private" cloud. This is a great step in maximizing the current capital investment in existing hardware. The long term argument is, since the "public" cloud is multi-tenant (i.e. server capacity is shared across multiple accounts) the vendors can arguably provide much cheaper computing power by maximizing server utilization across their entire user base.
Here's an analogy: When electricity first became mainstream, every large building had their own mini-power plant. Generators were in the basement of every building, coal was delivered to the building, and electricity was generated for that building. Utility companies later came along and created a central power plant for the city and provided a grid for feeding power to these buildings. Initially, certain individuals resisted, wanting to maintain control of their own generators. But ultimately, economics won out. It was just cheaper to centralize this processing. I see the same transition happening here, but it will take time.
This article is meant to provide some initial steps to ease that transition.
Larry brings up a great point as well. The Cloud is a different infrastructure configuration and not apples-to-apples with whatever is in the local data center. Therefore certain types of testing (i.e. testing with integration of external components and performance testing) may best be done locally if the production app will ultimately be deployed locally. Other types of functional testing however (i.e. system testing, UAT testing), could definitely be performed in the cloud. In my experience, these are the bulk of the testing cycles that need to be done as part of a software development initiative. This allows you to keep a much smaller "shared" testing environment locally, for these specific types of tests, while the cloud can be leveraged for everything else. The cost savings would still be significant. If an app is ever fully deployed to the cloud in production, the dependency on the local data center goes away entirely.
Post a Comment