The rush to cloud computing is taking off, but why? Doesn’t anyone remember time sharing back in the early 70s when computers were just too expensive to own and maintain? Are the issues back then no longer with us, like bandwidth, security, and predictable costs? Computer processing cycles are a fraction of what they were 30 years ago so why are the cloud vendors becoming so numerous and so popular? Are they taking advantage of the fact they many companies may have no idea what processing, monitoring and maintenance costs really are and cloud computing will somehow improve the bottom line?
Cloud computing vendors do have a “story” for all concerns and in some cases they are right on. However, does the story fit your situation? Not all applications are created equal. A call center with thousands of end users located in one location cannot be compared to an online store where end users in the millions are “everywhere”. Each application uses a complex combination of web and back end database services, but the entry points are anything but similar. Both of these systems require high availability and disaster recovery, which automatically is flagged as costly by most IT budgeting managers. Good candidate for the cloud, right? What concerns need to be addressed?
Concern #1 – Responsiveness/Reliability
San Francisco and New York City are extremely concentrated work and entertainment centers. They offer just about anything you could ever imagine, except easy, quick and cheap access. Tunnels and bridges built more than half a century ago limit getting in and out of those cities. Mega infrastructure providers like Amazon, Google and many others including Microsoft are apparently extremely reliable but access is limited to the connection speed at the end user’s location. For millions of remote users using the internet, the "bridges" are many and fast. For the corporate commuter, the bridge is limited by simultaneous users (i.e., cars and lanes of traffic). Collisions and stalls are a possibility.
Will your application requirements withstand the bottleneck? Will the money saved in infrastructure hardware be outweighed by inefficiencies in call center agents navigating from page to page? Will you even be able to measure these statistics?
Concern #2 – Cost
It wasn’t that long ago there seemed to be a server or two per application. Even the latest and greatest IP based phone systems were now running on Intel based platforms and companies purchased servers a dozen at a time. A new application called for redundant production devices, and then there were staging, beta, testing and development servers. A new Microsoft release most likely required a couple of extra lab machines in order to get the IT staff up to speed. Capital expenditures were eating away at the bottom line not to mention the added utility costs in keep these devices powered and cooled. The time was ripe for virtualization and VMWare hit the nail on the head. The bare metal server technology introduced around 2005 allowed companies to consolidate multiple hardware devices into a single device with multiple virtual servers. The need to purchase dozens of “boxes” each time a new application was introduced was literally gone.
That was the good news.
The bad news was training and coming up to speed on this new technology was not going to come cheap, but anyone with a vision could see that the long term benefits were going to be worth it. This of course assumed no “short cuts” were taken in the transition to the virtual environment, like monitoring resources and disaster recovery. Those that did not make the conversion for one reason or another may see the cloud as their savior. Even virtualization implementers may see the added complexity too much for their IT budget.
Moving the problem offsite does not make the problem go away, it just changes its location. You still need to monitor your applications, and you still need to justify costs. Capital Expenditures that were written off and buried in a company’s assets will be replaced with invoices for services rendered in terms of CPU speeds and gigabytes per month. How will those numbers be charged off to various departments? This sounds a lot like timesharing back in the seventies. If you do not know how to do charge backs now, it could get worse.
Concern #3 – All or Nothing
Concern #4 – In-House can provide the Same Service
Can they? Does IT have the resources? A new online store accessible to millions 24x7 can be built, but when? Unless you already have the “big pipes” and a 24x7x365 operation, you are looking at a considerable investment, not to mention the time required to build such a facility. Virtualization is not going to solve the bandwidth problem; however, a cloud vendor with mega Internet pipes could get you off the ground in a matter of weeks giving you time to analyze the benefits and risks of doing it yourself. Google, Amazon, and eBay did not start off with billion dollar data centers so why should you? This concern is obviously tied closely with costs mentioned above.
Do the math.
Concern #5 – Security and Privacy
Companies install multimillion dollar security systems to thwart intruders yet according to laptopthefts.org, 32% of all data breaches were the results of stolen laptops, cell phones or other portable media devices. Hacking only accounted for 14% of all breaches. Chances are the security level at a cloud vendor exceeds anything installed at your location.
Encryption is key to any application requiring “for your eyes only” protection; however, many applications today do not implement real time encryption because of the performance hit on the application. Encryption is often limited to login or credit card transactions, which is where hackers concentrate their efforts by spoofing many of the popular online banking or shopping sites within a bogus email. The location of the application (cloud or in-house) has little impact on security levels with this “problem” and the owner of this becomes the challenge of the application developer, not the infrastructure.
Summary
Cloud computing will change the way we think about new applications; however, it is not an automatic fit for saving money on existing infrastructures. It needs to be analyzed and thought through independent of vendor marketing brochures and promises. They are not in this business for the fun of it. They are there to make a profit just like you. They may be able to save you a bundle, but you need to do the calculations and make the final call, not them. Take your time and do it right.
Will cloud computing take over? Are you ready for it? It has a long way to go before it takes over.
About the Author
Rich Goddu has over 20 years of premier information technology experience in cost effectively growing IT infrastructures ranging from small start-ups to larger information technology based companies. He has spent the last ten years as Director of Network and Systems at TriNet HR Corporation in San Leandro, CA. increasing productivity by constantly analyzing technologies and processes that were either consuming systems, people or both. One included, but was not limited to, the elimination dozens of point to point data lines with an MPLS (“cloud”) solution allowing big pipes between data centers allowing real time disaster recovery. Currently Rich is specializing in Microsoft virtualization and the cloud computing space as the need for 99.999% uptime becomes an even greater challenge as power and floor space costs are eating away at the corporate bottom line.
Rich received his Bachelor’s degree in Electrical Engineering from the University of Arizona.
No comments:
Post a Comment