As previously mentioned, the company is headed towards a domain server and an SQL server after 20+ years of workgroup practice.

I predate the virtualization age and I keep hearing about it from the hardware vendors. Must catch up ASAP.
I dread the word from struggling all my life with unexpected power failures, blackouts andunreliable cabling and internet service that are simply put, our say of life.
No redundancy will protect us from any of this shit and it existed it would be uno tainable and unaffordable for us.

Please point me in the right direction to the dumies section of the library

0-0

I would go over to Pluralsight Training site and sign up and get what you can for free, sign on for longer (pay) if it helps. I have used it quite a few times.

https://www.pluralsight.com/br...t-ops/virtualization

I use MS Hyper V but there are others VMWare,etc.

Also LOTS on Youtube but I will let you pick out which one works for you.

Some other good links:

https://hyperv.veeam.com/blog/...-hyper-v-technology/

https://technet.microsoft.com/...831531(v=ws.11).aspx

Let us know if you have any questions.

One more thing ~ virtualization in of itself in not synonymous with redundancy. However, configured in a cluster situation it can be.

First of all, virtualization is really a wonderful thing from a system administrator's perspective. The ability to "snapshot" a guest system before making changes/doing upgrades, being able to roll out new/experimental server platforms pretty much at will, etc. gives huge freedom to operational decision making. It does cost real hardware money to set up a virtualization environment, but it is cheaper in the long run than buying a bunch of single-purpose servers and having to manage them. But with your eggs all in one basket, you better be damned sure that basket is built the way you want.

There are multiple parts to virtualization. The vendors want you to think they are all one and the same, but (for some part) they are mix and match.

- The host OS/monitor (ESXi, xen, bhyve, KVM, VirtualBox, QEMU, etc.)

- The management software for the above (VMWare/VSphere, OpenStack, HyperV, iohyve, etc.)

- The storage component and how it is presented to the management environment (block/file systems/etc.)

- The range of guest OS's supported by various combos of the above.

- hardware resources available to run the above

Work from the bottom to the top, if you are MS all they way, and don't care about running other OS's, the MS "stack" (meaning HyperV) is going to be hard to beat, especially if you are looking at fewer than 20 guests.

If you have to support a range of guest OS's, or lots of guests, the choice is much more difficult, with the devil in the details.

VMware over ESXi has, by far, the most widely-supported and readily available turn-key "product." You're not going to get fired for recommending VMWare (although you may go broke keeping up with the licensing requirements).

AWS, Rackspace, and others who have massive investment in this architecture use xen and OpenStack, with home-baked proprietary tools to make server roll out and management web-easy. In-house, you don't need these tools, and there is a steeper learning curve than the above, but if you are looking to roll out the "AWS for Argentina," this is where you'd start.

For most small to medium sized environments, say 10 to 20 production servers, and perhaps the same number of "experiments," not wholly Microsoft, where license costs are a limiting factor, I'd look at VirtualBox/Linux, KVM/Linux, and bhyve/FreeBSD. If you have money to spend, VMWare.

The storage component is as important as the actual VM "stack," again depending on the guest OS's and VM management software you intend to run. This can be a SAN, host resources, or something like CEPH. Make sure there is enough (and then double it), that it is quick enough, and that it is supported by all the players.

Rereading what I just wrote, this is a serious over-simplification of the subject, there are lots and lots of "gotchas" in this space. But it really isn't hard to get a basic VM environment going, assuming that the hardware base is there.

this was not my project but network administrator bragged about stopping ransom ware a couple of years ago with using the full version of VMWare. He was using the snap shoots option to do backups.

am a believer in VMWare.

you can also get VMWare player for testing OS's. this will run on a computer and let you make and test virtual OS's.

Good luck and have fun.

Bill

Don't want to get way over my head on my first try but this is roughly my scenario:

Buying two twin HP servers, hopefully DL360 with 32GB RAM and mirrored 300GB SAS boot disks and 4x900 GB SAS drives.
One server will run dedicated to a warehouse manager SQL solution and it provided by a company run by a bunch of unruly clowns that won ' t even provide me with a brochure of their product and services. These bozos are contracted by another company consultant that petendstoplay Administrator but should not be let alone with an isolated DOS machinecause he will hurt himself.
Other server will be my domain server.
Clowns and bozo frighten me more than technical challenges.

0-0

You may want to look into Microsoft or Amazon cloud.

Let them worry about the hardware.

Might be much cheaper in the long and even the short run.

0-0, IMO, your architecture makes no sense. Specifically, "dedicating" one system to the SQL server function. You are wasting half your potential VM resources.

Make both machines VM hosts, run multiple guest servers on them, one of them the SQL server, one the AD, a third the management/utility machine, a server to run backups, additional servers as needed for the components (e.g. Exchange) that will undoubtedly be needed. Try not to run multiple apps on a single guest, roll an additional guest instead. Spread the guests across the two machines as resources dictate, making sure to leave a little breathing room.

One of the hosts will get the 4x900 GB array, this can be 1+0, 5 with a hot spare, or 6, but any of these ways you net only 1.8GB from it. I'd highly recommend that you use bigger SATA disks for this, perhaps 4x4TB. Make sure the drives are rated for 24x7 operation.

Tie the two hosts together with a dedicated GigE or better. No other devices on this LAN.

Now, I'm not an MS guy, I'm Unix all the way, but this is how virtually every Windows VM environment I've observed is set up.

Pick either VMWare or HyperV as your virtualization environment. VMWare will have more and better learning resources, but at a greater retail cost.

And, actually, if you can do it, as sig2392 noted, the cloud option is becoming increasingly attractive.

I would start with the work station version of VMWare, http://www.vmware.com/products...rpro-evaluation.html

learn what you can do with virtual OS. the virtual OS created here could even be copied to a dedicated VMWare server in the future. if you don't like a OS build just delete OS. You will be able to clone OS's you like. This is a free download.

there is a free esxi download that is a light version of the server also a great start. https://my.vmware.com/web/vmwa...ownloadGroup=ESXI600

with virtual server you can have many OS's on one server box.

as you turn other loos on the sql server, first make a clone. when they blow up the active OS, delete the bad and replace with a clone. Restore the data base from backup. your good to go.

Have fun.

Bill

As a beginner to virtualization, do yourself a favor and download a copy of VMware Workstation Player. There are many options available for setting up a virtualization environment, but VMware Workstation Player is easy to use, it works in "simple" Windows and Linux environments, and it's free for non-commercial use.

It doesn't do everything, and it's a long ways away from the enterprise-level virtualization solutions from VMware, Amazon Web Services, Microsoft, etc. However, it will let you set up some virtual machines, and that is what you need to get started.
Focus on the basics -- how to set up a VM, how to perform virtual power-on, power-off, reset, etc., how OS installation works in virtual machines, how storage is handled in a virtual environment, etc. It's pretty easy to understand once you do it, but actual practical experience is very important.

Once you get some virtual machines running on your own, you'll better understand just what it is you really want and/or really need.

By the way, in the spirit of the moment I too will put on my (former) architect's hat (because everybody else is doing it). In a setting where you have unruly clowns that won't tell you what they are doing, you have no business buying any hardware whatsoever, especially something with a specific memory and disk configuration like you describe. If they are going to be the administrator(s), then you need to demand from them their exact requirements of memory and disk space. This is the very sort of basic info that you would need to set up virtual machines, so it's good to get it now.

Now to be realistic, lots of environments are set up the way you are describing -- buy some vaguely-spec'ed hardware and tell some anonymous vendor to have fun. It might work for you, and it might not. The hardware specs as stated are not unreasonable for a data warehouse setup. However, it's a good idea for you to get real numbers on what your applications actually NEED. Part of virtualization is effective resource management, and it's better for you if you can start implementing best practices right now.

quote:

Originally posted by spexet:
By the way, in the spirit of the moment I too will put on my (former) architect's hat (because everybody else is doing it). In a setting where you have unruly clowns that won't tell you what they are doing, you have no business buying any hardware whatsoever, especially something with a specific memory and disk configuration like you describe. If they are going to be the administrator(s), then you need to demand from them their exact requirements of memory and disk space. This is the very sort of basic info that you would need to set up virtual machines, so it's good to get it now.

Now to be realistic, lots of environments are set up the way you are describing -- buy some vaguely-spec'ed hardware and tell some anonymous vendor to have fun. It might work for you, and it might not. The hardware specs as stated are not unreasonable for a data warehouse setup. However, it's a good idea for you to get real numbers on what your applications actually NEED. Part of virtualization is effective resource management, and it's better for you if you can start implementing best practices right now.

Whole agree with you BUT here are my facts:
The Data Warehouse Management Software company crew only demanded a dedicated server with basic requirements of memory and space (for a W2000 server). No further explanation or details; the thing will allegedly grow from there to infinity.
Currently running a Test / Developpment Server under WServer 2012 and when I originally presented them with SQL Server 2012 they got their panties in a bunch and demanded SQL Server 2008 be immediately brought in as replacement. That got MY panties in a bunch, since I fail to see myself giving the OK to a project that starts with obsolete software.
Since they want to be left alone and play in their box, good for them.
Their specs will be 32GB RAM, mirrored boot disks and mirrored DB disks + one disk for logs and such and another disk for whatever.

I figure I might learn virtualization on the other box and eventually get their stuff to work in a virtualized environment once it is operational and have it as a mirrored server.

Electrical power is a major concern for me, that`s where the word virtualization sort of scares me.

0-0

The requirements you mention are pretty vague, so hopefully this environment you are setting up is just a testing area (sandbox) for the vendor to get their act together. Assuming you have access to the system while the work is going on, one of the things you can do during setup and testing is to examine the system and get some basic estimates of what resources are actually being used. There's no need to get fancy yet -- review what and how much disk space they are really using, and fire up the task manager and see what they are really running.

In the meantime, you can get your own act together regarding things like electrical power and stuff. It is right to worry about it -- in production settings, a typical setup for virtual environments often includes redundant disk, redundant servers, redundant network connections and backup power (UPS, backup generators and sometimes even backup telecom and power utilities). The bigger items like backup generators and redundant utilities might be overkill in your situation, but if you don't have a UPS you might consider including one in the budget. If the power goes out it won't just be one "server" that goes down -- it will be all the virtual servers that depend on that power, going down at once.

Not unusual during summer to lose power for more than a day. BUT summer is the main season for the company since it is a cold storage business.

0-0

quote:

DL360 with 32GB RAM

What CPU are you using?
E3 Xeon chipset max out most of the time at 32GB and 4 core 8 thread.
E5 Xeon has more mem capability and increased cores/threads.
Virtualization likes lots of CPU's (cores/threads) and lots of memory.
Of course the intended use will determine how much.
The VM's themselves use less than a hardware only machine.

Additionally what is your reasoning for a "mirrored" aka clustered server?
That is a whole 'nother step up in complexity.
It is very useful if your environment always has to be "up".

I would either start playing around with MS HyperV or start looking into Vmware. A few good books are "VMWare vSphere Design" by Scott Lowe, or something like "Mastering VMWare Vsphere 5.5" aslo by Scott Lowe. The official VMware cert guide books are good as well.

I would at least grab a copy of VMWare esxi and start playing with it on one of those new servers. It can run stand alone for free. You can install it directly onto a 4GB SD card. You want an actual type 1 hypervisor, not Vmware workstation or any "Type 2" hypervisor. That's for playing around, not production stuff.

Depending on how many servers you will end up with, you might study up on "Hyper Converged Vmware". That's where you can have three nodes with local storage but it acts as one large redundant san and you can vmotion machines from one host to another without downtime.

The ideal way though, is a fast quality SAN used as shared storage between multiple nodes.

I do strictly VMWare, currently have 24 hosts in the production cluster with 1194GHz CPU, 6.75TB ram, and 133 TB of storage in use, running between 400-500 virtual servers.

I would recommend finding an IT vendor that you trust, and have them set this up for you.

20 years ago, when virtualization first came around, it was a much less complex technology, and easy to get up to speed on. Staying up to speed now has also made transitions that much easier.

Starting from the ground, today, and having to pick up 20 years' worth of knowledge? To complete a project that you are likely to do once in your career? Not a wise use of resources, IMO. Things like this are the very reason that IT consultancies exist. They do this every day, they are very good at it.

Mistakes made in hardware selection, hypervisor selection (or both), in conjunction with hypervisor misconfiguration can be crippling to an organization, and can be quite costly to rectify.

An ounce of prevention is worth a pound of cure.

Your company is in the cold storage business and very likely has backup generators. If you are in the same physical plant, get a UPS and a tap into the generator backup. The UPS will carry the computers until the generators come on line. Powering two physical servers and network gear can be accomplished with a UPS that draws from generator or street power.

As for virtualization: I see you are in Argentina, check with local sources to determine what is available in your location.

If you will be running Windows exclusively, Hyper-V may be a better bet.

Virtualization provides quite a few capabilities and may be overwhelming. Start with the basics. At the very least your can use virtualization to run your primary domain controller on one physical box and your backup domain controller on the other. Running SQL in a similar fashion requires additional software/features such as Always On.

Virtualization has been around for approaching 50 years, VMware has been a force for 20 years so there is a large body of information available. Leverage your vendor(s) to gain access to the support community.

quote:

Starting from the ground, today, and having to pick up 20 years' worth of knowledge?

Well you also get to bypass twenty years of mistakes and benefit from twenty years of hardware and software advancement.

Hello- we're all piling in with info, but it looks like he has 2x DL360's with local storage only.

Without a SAN, we should not be advising vSphere. vSphere and shared storage would be a great leap ahead of what can be done with 2 standalone physical servers, however we are not at that stage.

Free ESXi would be the route to go for now. It cannot easily be updated in place until it is a paid licensed product. The management tools are good, and intuitive, and would allow you to maximize your hardware.

I would get power sorted, then start looking at storage and vSphere.

I am not 100% behind MS Hyper-V, though I do like it.

I work with VMware fairly often, have my VCP6 DCV, and designed and built a 2-node VMware environment for a CRM app that supports 1000 users daily. this environment has 2012 R2 OS's SQL 2012, and multiple load balanced 2012 R2 RDS servers. I have consulted for more than a handful of VMware, XenServer and Hyper-V environments as well.

quote:

Electrical power is a major concern for me, that`s where the word virtualization sort of scares me.

Your power goes out and you don't have a proper UPS, it doesn't matter if it's physical or virtual. Also - if you are going to have a domain controller - you should really have a second domain controller as well.