I've got a client who's needed new servers for a quite while
Said new servers will be running Windows..."modern" Windows. I've been dragging my feet like they've got cinder blocks attached.
Sales stats for the server market are out for the first three months of the year and they are bad – bad meaning bad, not bad meaning good. According to Gartner’s estimations for calendar Q1, vendors’ global revenues slipped 4.5 per cent to $12.5bn, and shipments were 4.2 per cent lower than the prior year period at 2,601,600 …
We just did the same - with our move to the cloud we only need one primary server and we went 2016. I have to say, it went in without a hitch - updates & all.
If you don't like the look of Windows, fair enough, but in terms of a painless installation & configuration, it certainly ticks the boxes.
I wish the place I worked would. I checked Supermicro and their retail prices are far less but management won't go for it because they don't offer 3 hour onside like Dell does.
Sadly, my explanation that everything runs in a VM with hot failover so we can just buy an extra server so a dead machine can be down all month while still being cheaper, go on deaf ears.
Supermicro is cheap primarily because their quality control is poor and their board designs are hasty. All they do is follow every new platform/chipset release from Intel, AMD, NVIDIA, etc., then slap together a product that just barely manages to utilize these new bells and whistles without exploding.
Yes, their boards do explode into flames every once in a while.
If you want to have the bleeding edge latest kit, knock yourself out with some Supermicro. But plan on buying lots of spares and spending a lot of time on a troubleshooting bench.
Supermicro is cheap primarily because their quality control is poor and their board designs are hasty.
At least in our experience (and we've been through several hundred Supermicro systems, some of them retired by now due to old age - so I claim at least some expertise), their design and manufacturing quality has been good to excellent. We had no early mortality incidents and no long-term problems so far. We did have a thermal issue with a system cramming several high-end GPUs in a 2U chassis; it was eventually replaced by a functionally equivalent, but more expensive set of two systems, at Supermicro's cost.
The worst systems we ever had to deal with were in fact from HP - one of their basic "HPC" models, I forget which one - horrible case design (you had to disassemble nearly the entire system to replace anything at all - even one of the case fans - and with none of the Dell's tools-free finess), horrible manufacturing quality (rasor-sharp edges everywhere; if you were servicing multiple systems rapidly, cut-proof gloves were a must), essentialy no burn-in or QA prior to delivery (on that particular batch we had infant mortality of over 25%, with CPUs, memory, disks, motherboards, fans, PSUs failing at random).
After 15 years of buying Dell/Fujitsu/HP etc servers, this year we switched to self-assembled Supermicro kit. I suspect we are not alone.
We switched to Supermicro a while ago. Never even looked back - their quality is at least as good as Dell's, at a far lower price, and having one box down for RTD service out of a hundred (which does not happed often, either) is not an issue for us. We were always taking RTD even when we were still buying from Dell, HP, and IBM.
Dell's growing because they understand automation and management.
You may or may not save a few bucks on Supermicro, but your business will incur bigger costs on the lack of automation, lack of management integration and all the pieces and parts that you now have to fuss with, unless you work for free, or you're a fixed cost whether you work 30 or 60 hours.
What if you want to roll out over a period of time? Dell maintains standard components. Ditto with a common firmware across multiple form factors. Supermicro can't do either of these.
My experience is that outcomes are far better for customers choosing Dell over other products. Supermicro customers struggle more than Dell, HP and Cisco and have higher management and maintenance burdens.
AC - because I work for a company that positions and sells multiple server products
AC - because I work for a company that positions and sells multiple server products
.. which the rest of us usually call "Dell". If we are being polite[1] :-)
[1] I actally don't mind Dell. Although, when I was buying servers, their tech support was in Ireland and very helpful and knowledgeable. Whether that's still true I can't say.
While a few industries might actually need latest/greatest, for the most part there has not been any compelling technology out there that makes a 2017 server much better than a 2013 server. Support costs will go up after 3 years but not to the point that refreshing and the new deployment services are a good deal.
With virtualization and multi core/multi socket hardware you can cram a lot of workload onto a few units...so, one server today might equal several servers a few years ago. There go the "Units shipped" numbers. While the heavy duty virtualization optimized host will cost more than one built to run one discrete workload, it will not be enough to offset that one server can do the work of 10 or more a few years ago.
Dell worked that math out long ago.
Dell also has a OEM business that sells server hardware for appliances, boosting their numbers relative to the other companies that have not figured that business out.
So far as Supermicro, the support model really isn't designed for a regular IT shop at all. The hardware is OK, but the business model is much more that of an ODM.
My career has seen Mainframes to minis to physical x86 servers to racked x86 physical servers with attached storage to physical with SAN storage to blades to VMs and now I am working daily with AWS and SoftLayer cloud.
We are shutting down hardware, consolidating their workloads, pushing what we can out to cloud and scrapping the old hardware, not buying them.
There's your issue for the smaller end of the market. It's all fine and dandy saying O365 is the way to go, but for smaller operations with small ADSL pipes, O365 is not an option and the cost of the Licenses for an on prem server of 2016 is unjustifiable (especially if you can still get 2012R2).
Dare say the cloud lovers will downvote me, but I will supply what the customer wants, not what a supplier thinks I need.
Are you sure?
O365 is what you want it to be. Nothing but a way to manage licenses; Exchange aaS; all the apps; online only; online and offline.
Having a small pipe to the internet will still cause you dramas with on prem Exchange/SharePoint/Skype.
You want to stay on prem? Knock yourself out! But the reasons (to go to the cloud) are usually commercial, not technical.