The Avalanche of Cloud Computing

  • Like anything else, there are pros and cons.  My biggest concerns are that security is a HUGE issue, and the lack of control of the product.

    Example 1:

    People like to think that the cloud is more secure. This is fundamentally impossible.  Every system has vulnerabilities. In an owned data center you have whatever weakness you have.  Add in the cloud, and you now have those weaknesses.  Since statistically it is unlikely that you have the same weaknesses, you can only increase the number, not decrease it. Why? Because I have to be able to use the system in the cloud, which means I have to be able to get there from my corporate network, which means that anyone that can compromise my network is automatically "in" the system in the cloud.  The assumption being that someone has compromised an account that access the cloud data, if we ignore those, then this discussion is irrelevant.

    Example 2:

    MS O365 performance sucks compared to locally installed Office, and every couple of weeks our software is updated with "improvements" that decrease productivity significantly.  Our finance team has some of the brightest people I know when it comes to Excel, and they hate it due to having to relearn things so often.  I have had things like filtering disabled due to how Excel imported data - which was NOT an issue a month ago. Can I do something differently to fix this, sure, but why should we have to relearn a tool that we already know how to use and are happy with, just because the vendor wants to do monthly "feature" updates.

    Performance wise, O365 has a known issue which requires that you limit your RDP session display settings, and that only has an incremental benefit. If I use the best display settings I can type a paragraph, and then watch as the characters appear on the screen.

    Summary:

    There are understandably benefits to hosting, specifically things like offering small companies the ability to get servers up quickly among others.  The cons are just as important, if not more so, and hosting providers are not addressing those concerns quickly enough.

    Dave

  • Mike Dominick - Thursday, September 21, 2017 7:53 AM

    I would say the biggest change that could create a shift back to on-premise will be bandwith costs.  

    The Telco's are sitting back and waiting for the tipping point where they know all your data is sitting in the cloud and the only way for companies to access is it is through them.

    At the moment companies gets entrenched in the cloud (Both client and vendor) the Telco's will be in position to cash in.
    They will be the modern day railroads and as of right now there is no way to move data without them.

    They will attempt to move to a new model where companies are charged on a per MB to use their data and for priority access. Vendors will also get charged on the other side for priority access.
    We already this happening today when the telco start throttling home connections or connections to services like Netflix.

    If that starts to occur cloud adoption will slow, especially for centralized companies, and could also be the trigger to consider re-centralizing staff back into the corporate office.

    Nice!  That is so obvious I have no idea why I didn't think of that.  The sad thing is you can't control it, because it doesn't matter how much you host, as other companies do, your internet costs will go up regardless.

    Dave

  • I've noticed a trend that occurs in new inventions.
    Generation One: We try and use it in almost the same way as the predecessor
    Generation Two: Some form of interim state where we realise that we don't have to do things the same way but have not yet realised all the possibilities
    Generation X: We use the technology in the radical new way that releases its full potential.

    I think we are near the end of Generation Two for cloud applications.  There's some really exciting stuff happening with cloud data warehousing.  You need the storage to be highly durable but your compute can flex dramatically.  There's huge efficiency and cost savings to be made provided the engineering is done correctly.  In a physical data centre you buy hardware sufficient capacity to last for the time for the costs to be fully deprecated.  That means swag estimates for peak load in 3-5 years time.  In the cloud you can buy for todays needs and pay extra if you need it.  One infrastructure colleague told me that the most attractive scaling feature was the ability to scale down rather than up!

  • Steve Jones - SSC Editor - Thursday, September 21, 2017 9:13 AM

    bdcoder - Thursday, September 21, 2017 7:49 AM

    Of all the words used in computing, the one I hate the most is "Cloud"! When did a warehouse of computers suddenly become a "cloud"?  Clearly, I missed that marketing meeting!

    If you think that a cloud is a warehouse of computers, you haven't really thought about what a cloud is. A cloud of computers can be a bank of systems where one takes over for another. That's way beyond what a series of computers in your data center or a warehouse is. There is substantial infrastructure in Azure, AWS, even Rackspace that allow another system to take over for a failed item, without you doing anything. It's beyond what ESX does, though more of an evolution at scale.

    The platform  services are also more than a warehouse of computers. They are tied together in ways that handle load, scale, deal with failure, and even handle hardware disparities. That's not to mention the multi-tenant separation, threat detection and complex networking, and more.

    Calling this  a warehouse is more marketing than hyping the cloud.

    No, I think a warehouse of computers is a warehouse of computers and a cloud is water vapor in the sky;  I'm just not one that digests marketing gargon as easily as most; but in the end, who cares so long as the "infrastructure" does what it should 🙂

  • djackson 22568 - Thursday, September 21, 2017 11:53 AM

    xample 2:

    MS O365 performance sucks compared to locally installed Office, and every couple of weeks our software is updated with "improvements" that decrease productivity significantly.  Our finance team has some of the brightest people I know when it comes to Excel, and they hate it due to having to relearn things so often.  I have had things like filtering disabled due to how Excel imported data - which was NOT an issue a month ago. Can I do something differently to fix this, sure, but why should we have to relearn a tool that we already know how to use and are happy with, just because the vendor wants to do monthly "feature" updates.

    I've had issues where O365 Excel just hangs up and slows down to a crawl. Turns out that it's loading updates in background in the middle of the workday. WTF??

    The more Redmond "improves" things, the more I want to spend time on my Linux workstation...

  • Security is my main concern with cloud right now. The other would be the main one that many organizations have faced, which is many of these platforms are like candy stores to IT professionals and it's very easy to overspend because it's very easy to purchase whatever you want at the click of a button.

    But regardless of some of the cons, the pros for me outweigh them. I have switched to Azure from on-premise primarily to dive into the MPP offerings from my SMP system and I have enjoyed the switch thus far. It's still early in maturity and surely does some pretty amazing stuff that would be too difficult or impossible for my on-premise solution to do. I really enjoy the idea that I pretty much have a wide variety of services I can spin up and spin down to support my team for whatever data problem we run into.

    I can totally see these cloud platforms taking more of the on-premise market in the future just for their ease of use as well ability to scale with the business.

  • xsevensinzx - Friday, September 22, 2017 6:45 AM

    Security is my main concern with cloud right now. The other would be the main one that many organizations have faced, which is many of these platforms are like candy stores to IT professionals and it's very easy to overspend because it's very easy to purchase whatever you want at the click of a button.

    The issues we've found in addition to this are: -

    • It's easy to spin stuff up configured insecurely
    • You still need good infrastructure design skills.

    • You'll need really good security design skills

    • Do not skimp on training Do not skimp on training. Do not skimp on training....{repeat until it sinks in}  

    • Some thought needs to go into the structure and naming of AWS S3 buckets and file keys

    • Cost calculators should be called SWAG costers

    • To get the most out of whatever facilities you wish to use you do need training.  AWS Cloud Formation templates can be built to be highly parameterised and yet enforce disciplines

    • What's on offer, particularly within AWS, changes very quickly.  It is hard to keep up.  If your organisation is spending enough money to get a vendor account manager then it is worth getting time with the vendor specialists.

    I'd say that the cloud makes things a lot easier but you do need to be much more thorough in designing what you want and you will need a broader range of skills either personally or as part of your team.

Viewing 7 posts - 16 through 21 (of 21 total)

You must be logged in to reply to this topic. Login to reply