back to article What is the enterprise cloud?

The private clouds are coming. A few of them are already in place, lurking in the shadows, but in 2017 the Infrastructure Endgame Machines (IEMs) land and everyone starts being able to buy cloud-in-a-can. With private clouds moving along the hype cycle towards Commercial Off The Shelf (COTS) solutions, the race to rebrand the …

COMMENTS

This topic is closed for new posts.
  1. John Smith 19 Gold badge
    Thumb Up

    A nice summation of what it is and how it came to be.

    With some available tools listed as well. Thank you.

    The big picture, going from a lots of "best-of-breed" point solutions to management to more of an integrated solution.

    The danger of all "unified" approaches is of course how they cope with some piece of must-have kit they don't have a standard interface for (or possibly something with a proprietary UI it's makers really don't like you being able to hook into through an API).

    This time around it does sound quite promising.

    On demand growth of processing and storage when needed and (nearly) painless scaling up if that's still not sufficient. All the while quietly gathering performance data warning you of impending failure in time for you to retire the misbehaving bit of kit without anyone noticing. A smooth, dram free and affordable IT service.

    Time will tell how well the implementations deliver the concept.

  2. Nate Amsden

    not a useful article

    The article implies enterprise cloud is only needed at pretty big scale (the scale mentioned is pretty massive) - and for the most part I agree, many people who say they need cloud don't really understand the situation and are solving for issues that aren't really issues.

    The people at the larger scales that need this kind of private cloud don't benefit from an article like this.

    With the team I am on we are managing about 1,200 VMs and containers(about 30 hosts right now - more systems under management than all of my previous employers combined though less physical hosts), and there is some more private cloud initiatives though for the most part I think it is overkill, and people are starting to realize just how complex a problem it is to solve (just provisioning OSs and stuff is maybe 5-10% of the work). We haven't lost an VM or container since we moved out of public cloud about 5 and a half years ago.

    Our biggest application servers haven't needed to be scaled for 2 and a half years now(I overbuilt it using LXC back then to save on licensing costs(systems paid for themselves almost immediately) and it has just run ever since, high loads, low loads, no problems).

    The new application stack has to be scaled more they say they want to be able to scale to 10X - but still nobody has the useful information that can tell anyone how it can efficiently scale(throwing more VMs at the stack when the existing ones are running at under 20% is a bad idea to me). They are working on it though -- 2+ years after I started asking for them to. From the sounds of it, maybe many months before all of the data gathering is complete.

    Oh and I'll mention none of the infrastructure is converged. Hardware standpoint pretty much everything is manually configured still (exception is vmware host profiles which address some of the config). VMware is pretty basic - just enterprise+ and vCenter (both are 5.5), nothing else(from a vmware product standpoint). Though at this scale it really hasn't been an issue. I have asked for things like blades and infrastructure automation but to-date budget hasn't been there. Automation is happening higher up in the stack though. The underlying infrastructure has been so reliable I guess that there hasn't been any push to do much more with it. Everything basically fits in 4 and a half racks.

  3. Anonymous Coward
    Anonymous Coward

    The whole idea of cloud was and is not automation alone. You need automation, but the idea of cloud was also a utilization based IT model in which you could pay by the drink (minute) for IT resources and turn IT resources on and off nearly instantly. You are never going to be able to do that with private cloud. You still need to buy data centers, hardware, automation software, etc... and when you have purchased, you have a sunk cost which you can't get rid of whether you need it or not. Also, cloud was about the idea that on prem IT is underutilized. Some of that was just due to people running their environments poorly which could be fixed with better management, but much of it is just a short coming of the on prem model. You always need to build ahead of resource requirements with on prem. Most companies have tons of un or underutilized hardware because either A) seasonal or infrequent peaks which they need to scale to... have to scale for your one day a year peak on prem B) As it takes awhile to get a budget, order, receive, install, phase in new hardware (even in a clustered model), they have to build out for months if not years in advance which means hardware is sitting idle.

    1. Trevor_Pott Gold badge
      1. Anonymous Coward
        Anonymous Coward

        Re: Optional

        I actually read that article. "virtualization and cloud review"... there was a giant NetApp HCI ad on the side, obviously biased and funded by VMware, NetApp, etc... the on prem, old world tech companies who lose out if everyone moves to public cloud... but whatever. The article makes no sense.... The whole analogy of gas powered cars are to on prem IT what electric grid trolleys/car are to public cloud makes no sense. If you had a car/trolley tied to an electric wire grid, there would be all kinds of places you could not go... it would be a highly restrictive service. Whereas you can go anywhere with an untethered gas car. That is the opposite of on prem vs cloud. Cloud removes restrictions... you can spin up any workload under the sun in a minute on public cloud. You can have nearly infinite IT resources at any time. Cloud is faster, easier, allows for more options and flexibility. On prem IT is much more restrictive, slow and rigid.

        1. Trevor_Pott Gold badge

          Re: Optional

          Are you implying that I, as the author of that article on Virtualization and cloud review, owe or have some sort of allegiance to VMware or Netapp?

          HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAH

          You truly don't even know what you don't know. Considering the amount of shit I give those companies - especially Netapp - yeah...I can never take you seriously ever again. Mate, VaCR gives me free hand to write what I want. No checking with the editor about topics, no kowtowing to any vendors. I am free to eviscerate (or praise) them as need it.

          But hey, you keep on keeping on. Try the next house over. I think they might be willing to accept Amazon the Saviour into their hearts today.

          1. Anonymous Coward
            Anonymous Coward

            Re: Optional

            Not a big fan of Amazon right now, actually... more a fan of Google Cloud.

            AWS is regressing. Now they want people to buy reserved instances, if you want any sort of discount, which locks you down to a particular image for the length of the contract. It is the opposite of the way it is supposed to work. If people are buying reserved instances, there isn't much value in public cloud as you are locked down to a particular build in a particular data center... same as on prem. Maybe they spin up somewhat faster, but not much of an advantage. AWS's primary services, EC2 and S3, are kind of old hat tech too. They are getting better, Lambda, but not the focus.

            Google, on the other hand, builds the discounts into the service. So you pay by the minute and if you happen to have sustained usage over the course of the month for a workload, the price automatically starts falling. There are also way more serverless services, e.g. Cloud Functions, GAE, Spanner, BigQuery, etc, in GCP... which is the future. I don't want to manage VMs, cores/memory, network, firewalls, CDN, etc... it doesn't matter. Not interested in guesstimating capacity or resource requirements. I don't want to manually scale out and scale in a workload, even if it is elastic scaling. I just want to set service levels, e.g. round trip latency to PoP or, preferably, end user, and the cloud provider can figure out how best to make that happen with no manual intervention needed.

  4. Shameless Oracle Flack
    Mushroom

    Cloud automation at the right level is key

    "If IT infrastructure can be boiled down to an enterprise cloud, a desired state tool and git to version the configs, we've reached the point where burning it all down and restarting makes sense."

    Yes we have, but leverage cloud vendors like Oracle to handle the base level automation (for databases, storage, networking, and apps) and exploit Terraform, Chef, Puppet and other open source tools (fully-supported by Oracle) to manage your complete enterprise infrastructure, both on-prem and in the cloud.

    1. Trevor_Pott Gold badge

      Re: Cloud automation at the right level is key

      The only problem with what you've said is the word Oracle.

      And that's one hell of a really big problem.

  5. John Smith 19 Gold badge
    Unhappy

    You say "utility" I say Project MAC. You say "cloud" I say "Unknown jurisdiction data centre"

    And you say "browser," I say "universal dumb terminal."

    BTW "on demand" pricing for processors and main memory has been a feature of mainframe OS's (EG IBM and Univac) for decades. So you could say that in Mainframe land "Enterprise cloud" has been around for decades. IOW this is not the first time the IT industry has been around this particular block.

    As for "always available" the events at AWS East 1 data center should have been a wake up call that you have to make the effort to get any potential reliability improvement, it's not simple and you have to test it. Otherwise it's just like any other data center failure, with added lack of insight into what's going on.

    Cloud vendors, like all tech vendors, are keen on customer lock in. Anyone who hands their core systems data to such a vendor needs to realize they have lost a very large amount of control. While everything is humming along PHB's won't mind. CAPEX eliminated, low monthly charges etc.

    It's what happens when everything is not rosy that people realize just how hungry a fox is in the hen house and how little they can do about it.

    And I agree, once your dependent for even your processing cycles to an outside vendor I don't expect those low, low introductory price deals will last.

This topic is closed for new posts.

Other stories you might like