Win Server 2003 addict? Tick, tock: Your options are running out Windows XP is officially gone but its server companions Windows Server 2003 and Server 2003 R2 live – just not for much longer. Mainstream support for the server duo ended on 13 July 2010 but the expiration of extended support is now just three months away: 14 July 2015. The date is critical as that’s when security updates and …
Ok, so basically, you are saying move to Server 2008, to cloud, or both and keep a local NAS server running Linux. The BSD's/Solaris are MUCH better candidates for NAS services than Linux, ouch!!!
Cloud should be a NOGO, needles to say why ... if you read the register, you already know.
How about moving the server infrastructure to BSD completely and keep IT safely inhouse ? Samba/Muppet have come a loooong way, so much so that Muppet outperforms AD/GPO, even for windows clients. Throw Open/LibreOffice onto all clients, alongside MS office, IMAP email (to keep Outlook addicts happy) -> soft migration to free software.
In 2020, deploy Linux or Free/Open/DragonFlyBSD on select clients and phase-out Windows.
Does the author know IT ? cf icon
The author seems to have been concentrating on old must-have apps which don't migrate well to newer Windows. Moving them to Linux is an unlikely option. Going Linux is generally an unlikely option if someone has a single server environment, because there is unlikely to be IT skills available. I'm not saying Windows is easier, but if you already have it, you probably already know it, which is probably the most important thing.
"Hyper-V comes as part of Windows Server – making it a good choice for smaller setups."
Hyper-V Server though is a standalone and completely free product. No need for a Windows Server license.
Also a Windows Storage Server based NAS would be a better option for many than a Linux one - particularly if they have AD. Certainly it would like have more features and would for instance support DFS, compression and dedupe...
Cloud for most services will have a significantly lower long term TCO of course - where it is an option.
"I'm not saying Windows is easier"
It is - there is no realistic argument otherwise for the vast majority of use cases.
It looks you have no clue about what AD does in recent versions of Windows. Also, some good Windows applications do rely on AD schemas (maybe extending them as well) to work. Moreover, AD is fully integrated into Windows, while Puppet or other solutions requires add-ons to work (and you may have to pay for it). Sure, SAMBA got better, now instead of NT4 domains only it supports AD 2000 ones too! Just, without the management tools...
IMAP does just the email part of Outlook (and not everything). Printer management is a breeze with Windows, not so much with other systems.
And there's much more software than Office to migrate - I understand BSD/Linux users don't go beyond the little they found in their distros, and may be a suprise to find how many different applications people use on Windows.... and, unbelievable - they even pay for them without issues!
In 2020 Linux will still be on 1.46% of clients only, don't worry. But, sure, for a plain NAS I'd investigate a BSD one offering ZFS.
"The downside of migrating to a brand new Windows Domain is that any permissions set on things like file shares have to be redone"
Use ADMT to migrate user accounts and security groups with SID history and use robocopy to migrate files and folders, preserving permissions and attributes etc (DATSOU). Permissions are carried across properly this way. Use 2008 R2 functional level for your new domain and don't raise it until after you're done with ADMT. Easy.
I do think the author knows a thing or two about IT, whereas your approach suffers from a lack of understanding of several very basic facts about the whole business.
The article is aimed at people/setups which are firmly rooted in a microsoft environment, with the inclusion of the possible existence of some very specific business-critical business apllications developed for and on this microsoft platform. It's acually pretty comprehensive for a quick overview.
Now the thing with open source, as it has been from the early beginnings, is how well it supports your existing infrastructure when/if you decide to switch, and how much of the needed knowledge is already in-house if you would implement it. Which usually amount to double negatives for most small companies, and as such is as much of a No-No as moving your entire business to the Cloud.
Switching "Offices" alone gives a bunch of headaches, besides retraining staff, because the Microsoft variety is so well integrated compared to the Libre flavour. This means that you'd need to go over all of your scripts, letterheads, shared spreadsheets, etc. etc. to find out what gets broken ( and oh boy.... things will get broken.. ) and fix them. If Libre supports whatever-you're-used-to-doing at all to begin with. Libre has come a long way and is quite good nowadays, but it still has some idiosyncracies that makes it necessary to upend some things which are "natural" and expected when you're used to working with the Microsoft line.
Server-wise you're dealing with something similar. Your staff is trained in using/maintaining Microsoft servers. With all their little foibles and workarounds, and ...stuff... They may or may not be versed in Linux/BSD flavours, but even if so, they're not getting paid to administer them. They're getting paid for keeping the microsoft servers running, and their training and inclination will reflect that. They also keep those legacy applicaations running, and will in general have a dim view of some new kid coming in wielding the "Open Source is da Bomb!" hammer. BOFH's look for suitable PFY's.. The ones that aren't tend to meet the same fate as Sales Drones and Bosses. They generally are not waiting for a year-long session of More Work, More Headaches, More Incompatibilities forcibly introduced into their lives. More so if they actually have looked at Open Source in the past and encountered the flag-waving, barricade-bashing Lintard crowd and other types of cellardwellers frequenting the Land of Support in Open Source land.
Admittedly, these hurdles can be overcome. Technical issues are just that, issues that can be solved by the competent, and Software, open source or proprietery, can be assessed purely on it's merit and business case, ignoring the activists and trolls. Which leaves the business case. Because you will have to shell out extra cash to get things done, even if you choose "free" open source software. You will either have to get the knowledge in-house and keep it there, which means hiring new personnel, or rather ...replacing... the personnel you work with, or buy a service contract with one of the business oriented open source solutions. Which for small companies means you have to boot people you actually know and/or do business with a different kind of devil that so far has an unproven track record.
Whichever way you look at it, it will cost you, as a business owner, a lot of time, effort, money, and in extreme cases some Unpleasant Conversations to switch to open source. As opposed to having your existing personnel ironing out the few differences while upgrading to the next version of the devil you know. It's really a no-brainer, since the total costs of migrating to open source are simply far greater than extending/migrating within the microsoft environment.
IT, for a modern business, has indeed become part of the backbone. And just because of that, business owners and managers have become very conservative in applying major surgery on it, lest they break the spine..
A fine piece of well argued common sense. There is a case to be made for moving to open source but the "its free" argument is balls. As you rightly point out the economic case has to be treated with great caution and every business looking at these options has to ensure that they do not end up in deep how's-your-father because they did not understand where the bear-traps are.
>I do think the author knows a thing or two about IT...
Probably, but I suspect they haven't been through an SME upgrade from 2003 to SBS 2008 or even 2012. I've found if you approach it as a full IT transformation/refresh, you tend to lift the covers and discover all the stuff that will trip you up and cause the business to be without IT for a few nail-biting days...
Also in the SME space there is no equivalent to SBS beyond SBS 2011, so an upgrade to 2012 can be more involved, particularly if you are deploying Exchange, SQL-Server etc.
At a basic level, if you are running 2003 then you are probably well advised to buy a new server. But there are proven tools that will enable you to take a physical 2003 serve and spin it into a VM, which can then be run on practically any system running a modern CPU; useful if business continuity is a consideration.
There are people running 2003 on shiny new servers (or shiny new VMs), just because the latest time they upgraded hardware they didn't upgrade the OS - unlike clients when often a new hardware means also a new OS (but not so often in enterprises - see how long it took to migrate off XP), servers often get replaced when close to their EoL, but the software on them stays the same.
@LDS - whilst your point is totally valid, I was responding to the query raised by 'x 7' namely "OK....how do you do that without running the risk of nuking the existing server? Most people don't have a spare one hanging around"
Which implies smaller installs typical of SME's (ie. non-VM and non-datacenter) where typically there is no spare physical server nor MS licenses to cover such a server if it did exist... Personally, I purchased a 'recent' dual Xeon workstation/server off eBay for a couple of hundred and use it as part of my migration service to get around this problem...
The biggest problem I find tends to be the apps with 16 bit installers (It amazing how many apps we used which were supported by vendors that still used hideously old installer packages), my own experience is 32 bit apps run fine on 64 bit but non sub system to handle 16 bit apps anymore.
Lots of fun where companies are no more or when you ask them "will it run on such and such" the response is "We don't know, why don't you test and find out".
Its easy enough to unpack but takes time to find out what registry keys are needed, find out it has something like Interbase 3.x and then cry when they need to be given stupid security permissions because it wants to have a file hardcoded on somewhere like the root of the C:\
Having had to provide a migration path from SBS2003 for an SME my initial thoughts were Win2008/12 + Office365 except the exisiting asymetric DLS connection had insufficient bandwidth (an outgoing e-mail to your colleague at the next desk has to travel quite some distance out and back again - never mind skydrive for business synchronisation issues) and FTTC was not available. So, I looked at hosted desktops and it ended up ticking many boxes: significantly reduced support costs ( thin client bust? get the spare out of the cupboard), per user bandwidth requirements for rdp are quite low, backups are someone elses problem - as is disaster recovery (though both need to be tested - never were previously), Internet bandwidth is via the hosting company (gigabit), can now implement better spam filtering and web browser monitoring/filtering (costs money but worth the expenditure), MS Office now fully patched and 3rd party applications (Sage principally) are properly maintained. A hybrid solution can be implemented if there are applications on a local PC (Autocad is one, video of sound editing would be others) which chew CPU and/or are graphic intensive, where everthing else is moved off to the hosting company. Now have to dispose of many PCs and an old server.
Actually, I should have made two posts because I meant to convey two different ideas.
1. You want to have a NAS, use Linux???? <--- this is bad, hence to reference to "IT?" If you are gonna go the freetard route, go BSD for NAS, NOT Linux ... Linux is way better than Windows for NAS, BSD/Solaris, however, is in another league altogether
2. You consider moving IT data services to the cloud <-- bad, bad, bad idea, for obvious reasons.
Then, for some reason, auto-text kicked in with the "migrate to free software while you are at it" mantra ...
You will have to migrate away from those proprietary no longer maintained 32-bit software solutions sooner or later, why not now ? AND do not make the same mistake twice, go freetard - if it is good, there will always be at least one fork, should the freetard vendor go titsup.
Ooops, I did it again ... ;-)
I think I want more down-votes, yes, that's it ...
Linux is better for a NAS? Depends... EXT4 is not the best file system around, and got its flack too. RAID resilience is down to the RAID controllers and disks, not the OS (unless you use software RAID, but then you're going to build a really cheap NAS....)
From the FS system perspective, in many situations ZFS is a more interesting options. But in a NAS the file system is just a piece. Than you have the file sharing protocol, and the permission management. If you have to share a lot with Windows client (or even Apples...), you have to use SMB and Windows permissions (NFS is really a no go, sometimes even under Linux itself) - and while SAMBA improved a lot, still it's not a native implementation like in Windows, some pieces are still missing (check AD RMS, for example), and you have far less management tools to work with and integrate with the rest of Windows management.
Sure, if you have simple NAS needs, a few user/groups and simple security Linux is a cheaper alternative. ZFS is interesting because at least it could ensure a far better data protection than old RAID, and in some situations it could offset the "disadvantages" of runnin a *BDS like TrueNAS, but again, it depends on what you need to store in the NAS, and how you need to share it.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
