Recommended Posts

Don't need to know if Hyper-V is better or worse than ESXI or whatever other option there is :p

Ok, So I've shown the boss that we can upgrade newer software while keeping our current hardware by doing some virtualization.

NOTE - This is not a huge corp - LOL like 20 to 30 Users at MOST. I know we could do it all on Server 2012 Essentials or SBS 2011 but we need full fledge SQL for the new database we are gonna be running and SBS doesn't do Hyper-V I think. ESXI I guess could work there but still :p

We have 2 machines running Windows SBS 2003 R2 (#1) and Microsoft Windows Server 2003 R2 Enterprise (#2):

#1 Roles:

AD/DHCP/DNS/SharePoint(including SQL built in)

#2 Roles:

SQL 2005 for a program we use and File Sharing.

#1 runs on a Dell PowerEdge 2950 Dual (2) Quad Core Xeon 5405 - with 4GB now (Max is 32/64GB - can't figure out which!) 3x 15K SAS 146GB in RAID 5*

#2 runs on a Dell PowerEdge 2950 Single Quad Core Xeon 5430 - 8GB now. 3x SAS 15K 300GB in RAID 5*

* = I know RAID 5 sucks for write performance so that's already being switch to a RAID10.

Ok So what do you guys think about my plan to switch everything to #1, stuff some more RAM in there up to 20-30GB and virtualize?

So it would be like this:

#1 = Windows Server 2012 - AD/DHCP/DNS/File Shares + Program we use now but using SQL Express it comes with (has to be 2005!!)

#1.1 = Windows Server 2012 - Exchange 2010 and IIS for the web and ActiveSync stuff.

#1.2 = Windows Server 2012 - SQL 2008 SP3 + NEW Software for new database we getting.

#1 = 4 - 8GB Of RAM

#1.1 = 12GB of RAM

#1.2 = 8GB of RAM

Sounds do-able or am i just dreaming?

I am running #1 and 1.1 now on another (older) 2950 with 9GB and been part of the domain for like a week too and seems stable (other than known issues :p).

Any tweaks or changes you guys think would make it run better? Thanks!

Question 2 - Storage is an issue if you ask me, boss being cheap! I wanted to get 4 x 1TB SAS 7200RPM drives for $150 each to build the RAID 10. Now my plan is to use 2 x 146GB in RAID 1** and the 4 x 300GB 15K SAS in RAID 10.

So I know it's not recommended to run Exchange on a dyanamic VHD - but what if I ONLY put the Windows Install there but put the Information Store and other data that Exchange writes to (log and such) on a FIXED VHD of say 60GB while keeping the Windows/Exchange Install to a dynamic C drive to cut down on space usage (only like 20-30GB VHD then)?

** - Should I do RAID0 instead of RAID1 for the 146GB drives (OS ONLY, no Data) for more performance since we backup to a USB Drive + (PLUS) OFFSITE backup evernight? Shadow Copies - which they (offsite people) test by booting it in a VM to see if it works.

Edited by SHoTTa35
Added clarification to the RAID0 question.

Idk, esx is very easy to use, almost brainless even. The host os uses very little resources. I would certainly up your drive space. Do not do raid0, I would not recommend that practice in a business environment or on servers. You want the ability to fail and keep running.

Think of it like this, how critical is it for the business to stay running? If it can stop for a few days while you completely rebuild the servers then do RAID 0, if it can't afford to be down, do not do RAID 0. I don't know of any businesses that can be afford to be down for mutilple days. The best that I have seen for disaster recovery is 15 minutes up time, with configuration of the networks involved. The worst is your scenerio and what you are contemplating (RAID0 in a majority vm environment). If you implement your solution have the business owner prepared that he will never open his doors again after a drive failure.

You can sure virtualize everything to keep costs down on hardware and only consume software costs. You will need to Max out those boxes with memory (the free vmware hypervisor maxes out at 32GB). I would also try to either max out the drive space on the servers or I would consider purchasing the essentials plus from vmware or purchase the datacenter server from microsoft to be able to support iscsi and get a iscsi appliance to increase drive space exponentially.

Exchange 2010 doesn't currently work on server 2012, hopefully it will with a future update. Because of this I'm using Exchange 2013 Preview at home lol

Exchange 2010 does "work" on 2012 actually. I'm using now in a live production setup too (bad bad! I know, boss doesn't like "it's supposed to based on X" - wants to see it working before he commits to costs). - used this to get it installed:

http://syscomlab.blo...ws-server-2012/

Not to say there aren't a few little issues but those are from Upgrading from SBS 2003 R2 to Exchange 2010. It's not supported obviously but it is running now in a Hyper-V for testing - currently synced with the SBS 2003 R2 Exchange.

There's the can't install SP2 RU4 (error 1603) but that's not specific to my setup as people with 2008 R2 and 2010 having issues too.

/offtopic

How is 2013 Chris123NT - i would so do that too as i'm a beta guinea pig but guess i wouldn't put my job at risk for that :p 2010 is definitely a step up though from 2003, not sure how much diff there is with 2010 vs 2013 however. :)

post-698-0-12376700-1349975894_thumb.png

Idk, esx is very easy to use, almost brainless even. The host os uses very little resources. I would certainly up your drive space. Do not do raid0, I would not recommend that practice in a business environment or on servers. You want the ability to fail and keep running.

Think of it like this, how critical is it for the business to stay running? If it can stop for a few days while you completely rebuild the servers then do RAID 0, if it can't afford to be down, do not do RAID 0. I don't know of any businesses that can be afford to be down for mutilple days. The best that I have seen for dr is 15 minutes up time, with configuration of the networks involved. The worst is your scenerio what you are contemplating, if you implement your solution have the business owner prepared that he will never open his doors again after a drive failure.

AS for sc302 - I agree with you that's why i initially said RAID1 for the 2 x 146GBs (total 6 drives MAX). Internet being down is a bigger deal than server being down (even with the server doing DHCP/DNS, that's easy to bypass though). Email also being on that server so could be an issue but we use a spam appliance (Barracuda) that captures emails before delivering them to the server so email would still be received at aleast :)

What I was thinking is to install Hyper-V Server 2012 server on either of the other 2 Dell PE2950s we'll have left sitting around which can then run the VHDs from the Exchange and SQL. Those files would be on the RAID10 600GB Virtual drive (4 x 300GB) So those can survive failures plus the USB backup (3TB USB drive) plus Offsite.

You are right though, I was just hoping to have more storage. 4 x 1TB would give us 2TB total (reality, only using about 400GB now) so that should last a while.

They have this sweet sale now which to me made more sense than spending $120 vs 150 for a 300G vs a 1TB:

http://www.newegg.co...N82E16822148868

Wanted to get 4 of those.

raid 5 with a hotspare is fine that would give you 2 drives to loose. don't worry too much about the speed, data integrity is where it is at in business world.

Also remember when backing up, if you backup just data (which is fine) how long would it take you to bring up the environment to the point where it can receive that data? If you have to recreate all of the servers in a virutal server or a physical server from nothing it is going to take a whole lot of time to do so. Even images can take awhile if not kept up on. And services are not cheap for this type of dr. If you put all of your eggs in one basket and that basket breaks what then? Will you be able to put together another basket quick enough to allow the company to continue to funciton without loss? How are you backing up? What are you backing up? How are you going to recover from an absolute failure? What will you have in place to prevent absolute catrostrophic failure? With multiple servers just backing up data isn't enough, and having multiple servers in one basket is asking for trouble. Just like raid, you need redundancy on the server side or you need a powerful backup that can restore instantly.

Information noted (Y) :yes:

The backup consists of snapshots of the OS as well as Exchange/SQL data and data generated/collected for reports. We do lots of Word/PDF docs and reports and scans of paper to compile said reports. That's what's in the file share on the server. If the worse happen and say all HDDs die at the sametime, those drives could be replaced and then imaged and be back up and running (depending on how long the copy goes, probably a few hours for all data, less for just server and email. If say motherboard fried then a temp machine from the backup company can be brought in (local guys) and just mount the backups images we have and boot them.

Preventing failure - hmmm guess need to think about that part some more since everything does eventually fail. I was thinking of having the other servers around for backup hardware incase one should ever fail. As appraisers, if the computer goes down they can go do field work (collecting data) while the server gets put back together. Can even work for the most part without the information on the server, unless that is, the report is almost complete then gets lost because of a server crash! :(

Still wondering about the questions of dynamic VHDX files though for Exchange OS partition, still googling too. Don't tell me all you guys are installed to the hardware for each individual server? :D

I personally haven't made a thin provisioned db server. I would think that fragmentation would eventually kill u. I don't see why you couldn't do it for the is side being that it won't grow all that much.

You know it may be a whole lot cheaper to go exchange online. Ms is offering an unlimited space for $8 per user with their exchange online plan 2. IMO it is hard to ignore that, that includes spam and virus filtering.

Trust me, we went thru this with the boss. He said they had it few years back and just wasn't stable (probably their connection to the internet at the time too wasn't stable), that plus the reoccurring costs vs pay 1 time up front. I agree with you though, it's like $800/year ($4/month plan) vs tons money up front. Our internet has gone down maybe 2 times this year and it was more a widespread issue (NYC Region - fiber cut and "hardware failure") rather than just bad management. So it's not like the old days where they might have had DSL or something. It's a dead issue I guess... I was like we definitely don't need Exchange on premises, I think it's more the principle than anything. :p

I got the extra HDD from UPS today so just waiting on RAM now and I can start building her up so that's why I wanted to get a few questions knocked out :)

and with local caching, there really is no reason not to consider it. We have all been down the road where services have been severed, but that was before. Services have since improved greatly, and like you pointed out 400 or even 800 per year vs what could be easily a $10,000 install up front and the software will stay the same version until you decide to upgrade again with another 10,000 up front.

hardware + software + implemenation time.

Local caching should solve all those issues I agree, then again preaching to the choir here. :p

Basically got my extra 8GB of RAM now (already got the extra 300GB HDD to make a RAID10). Not doing full mode install just yet but gonna at least test out the RAM to make sure they are stable in one of the other machines.

So for the current setup, no other suggestions to for implementation? Seems good as I mentioned it?

Thanks again for helping me think about some things I might have missed :)

raid 5 with a hotspare is fine that would give you 2 drives to loose. don't worry too much about the speed, data integrity is where it is at in business world.

I disagree. If you are going to set up RAID-5 + HS, you might as well go with RAID-6.

Why?

RAID-5 with a hot swap means when a drive fails in the active array, it can immediatley begin a rebuild. Rebuilds degrade performance and have a chance of failing (URE).

RAID-6 means that when a drive fails in the active array, the rebuild process does not begin until a new drive is added to the array, and you'd still need to lose TWO more drives before there is data loss. Even being down one drive ("degraded") you could perform a solid backup incase the RAID-6 rebuild fails. Granted you could do that in RAID-5 in a degraded state, but you can't if it automatically starts rebuilding...

Welp doing it now. :s

Moved the SATA 160GB RAID5 drives to the new machine and imported the config and it booted. Gonna have to clone those drives now, RAID-5 SATA to the RAID1 SAS 146GB drives then boot that.

After that's all said and done it's a matter of creating the RAID10 4x300GB drives then copying the data back from the external drives to the server and we should be done :yes:

Just to copy all this data is what's taking forever :p

EDIT- 10/22/12

No complaints yet half-day into the new work week. Some folder redirection stuff didn't happen so smoothly but (Favorites/My Documents) but working those out slowly but surely.

Still having issues though with Mapped Drives - can't for the life of me get it to map (Local admin/UAC problem i hear). That is they still map via loginscripts that was used in 2003 days but new GPO mappings in 2008 and up isn't working. There is a problem on Windows 8 (client) where it wont work properly but i'm doing this on a 2012 server with Win7 clients. Gotta figure out that thing were you can see how they are applying and if anything has failed or whatever. :p

Edited by SHoTTa35
Adding updates
  • 3 months later...

Sorry to add this late into the discussions but you spoke of Hyper-V, you can download the Hyper-V 2012 Core RTM free of charge and run your VM's on that. I run a Hyper-V environment in a clustered configuration and it works really well. I would say the only downside is that ESXi supports Linux and other OS's better, but it sounds like you using an all Windows Environment anyway.

You can download the tools to run on your workstation (must be Win7/8).

Just offering another option, it's got quite a low overhead as well.

http://blogs.technet.com/b/keithmayer/archive/2012/09/07/getting-started-with-hyper-v-server-2012-hyperv-virtualization-itpro.aspx#.URGHFlqu9W0

  • 1 month later...

Sorry to add this late into the discussions but you spoke of Hyper-V, you can download the Hyper-V 2012 Core RTM free of charge and run your VM's on that. I run a Hyper-V environment in a clustered configuration and it works really well. I would say the only downside is that ESXi supports Linux and other OS's better, but it sounds like you using an all Windows Environment anyway.

You can download the tools to run on your workstation (must be Win7/8).

Just offering another option, it's got quite a low overhead as well.

http://blogs.technet...px#.URGHFlqu9W0

Hyper-V does an alright job of running linux...at least RHEL and ubuntu. I've got 4 Ubuntu VM's on a Hyper-V server and they run great. Only thing that sucks is no updated integration drivers for Ubuntu, so you can't use Dynamic Memory.

This topic is now closed to further replies.
  • Posts

    • Instagram now lets you manually reorder posts on your profile grid by David Uzondu Instagram is finally rolling out the ability to customize your feed layout as you see fit by letting you reorder posts on your profile grid. This feature comes several months after the app introduced a tool that lets users rearrange photos and videos within a carousel post after it has already been published. To do that, people tap the three-dot menu in the top right corner of the post, select the edit option, and reorganize their slides. Now that Instagram has expanded the feature to your profile grid, you can organize your main page without deleting old uploads. To use the new system, you simply tap any picture on your grid and select the option to reorder. This action opens up a separate screen where you can freely drag your grid items around until you get your preferred aesthetic, and then you just hit the back button to save your changes. Instagram's Threads account posted that the system would reach accounts starting this week, so you might need to wait for the automatic update to hit your phone. https://www.threads.com/@instagram/post/DZVV_fyjjSW In other Instagram news, last week, people figured out that if you ask Meta's AI support assistant to hand over any Instagram account, the bot will actually hand it over (even if the victim's account had 2FA enabled). The security exploit involved the assistant accepting prompts from users and generating password reset links for unauthorized email addresses. Meta said that the issue has now been fixed, but this came after the issue affected several high-profile accounts, including @obamawhitehouse. Last month, the company finally rolled out paid subscription tiers for WhatsApp and other Meta social platforms after months of testing. WhatsApp Plus costs $2.99 a month and gives you custom themes, while Instagram Plus and Facebook Plus cost $3.99 a month for extra profile customization and story rewatch counters. Meta's also working on Meta One, a unified subscription service that contains options for heavy users of its servers who want more reach or advanced features. For instance, Meta One Essential ($14.99/mo) comes with a verified badge and impersonation protection. If you pay for Meta One Premium ($19.99/mo), you get deeper AI reasoning tools, whereas the Meta One Advanced ($49.99/mo) tier increases your search placement (on Facebook and Instagram) and visibility.
    • Hello mysterious lamborghiniv10, I was in Australia and... now I'm in the Netherlands. 
    • EU says Meta must restore rival chatbots' access to WhatsApp by Hamid Ganji The European Commission has ordered Meta to restore third-party AI chatbots’ access to WhatsApp after the tech giant decided to block them from operating on the popular messaging platform. After Meta banned rival AI chatbots from operating on WhatsApp, the European Commission launched an antitrust investigation to determine whether the company had abused its market dominance. As a result of Meta’s decision, third-party AI chatbots, including Microsoft’s Copilot and OpenAI’s ChatGPT, were prevented from operating on WhatsApp. At the time, Meta said it wanted to reserve the WhatsApp Business API for other types of businesses and did not allow rival chatbots to use it. This effectively prevented the WhatsApp ecosystem from being used to distribute rival chatbot services. However, the European Commission has now announced an interim measures decision requiring Meta to restore access to WhatsApp for rival general-purpose AI assistants on the same terms and conditions as before October 15, 2025. The Commission has also asked Meta to maintain that access until the antitrust investigation is concluded. The Commission argues that Meta has used its dominant market position to prevent rival AI chatbots from accessing the WhatsApp Business API. While Meta allowed rival services to return to WhatsApp by paying a fee, the European Commission still considers that arrangement to be a de facto access ban. According to EU antitrust chief Teresa Ribera, the fees introduced by Meta are so high that using WhatsApp is no longer economically sustainable for competitors. “It seems that Meta expects to leverage the vast reach and likely dominance of WhatsApp to benefit its own AI assistant and to foreclose rivals,” Ribera said. “We cannot let large digital incumbents leverage their dominance of the past to dictate who in Europe gets to compete and who gets to innovate in AI.”
    • A few years ago walmart had the 512 models on clearance for $35. I bought 3 of them. I should have purchased more.
    • I'm fine with a little reasonable promotion of Edge, but the degree which they do it right now I consider extremely unreasonable. 
  • Recent Achievements

    • One Year In
      Primer1st earned a badge
      One Year In
    • Experienced
      JayZJay went up a rank
      Experienced
    • Reacting Well
      Sir_Timbit earned a badge
      Reacting Well
    • Week One Done
      rubentuben8 earned a badge
      Week One Done
    • Week One Done
      ARaclen earned a badge
      Week One Done
  • Popular Contributors

    1. 1
      +primortal
      512
    2. 2
      PsYcHoKiLLa
      229
    3. 3
      Edouard
      134
    4. 4
      ATLien_0
      87
    5. 5
      Steven P.
      80
  • Tell a friend

    Love Neowin? Tell a friend!