Should I be concerned? Exchange 2010 config

Soldato
Joined
18 Oct 2002
Posts
7,622
Location
SX, unfortunately
I think this is a rhetorical question... Am now overseeing the company IT instead of just being on the fringe. I've spent a few hours seeing what I have. I've fixed a large number of DNS and DHCP issues that date back to when the main servers were 2003 and were upgraded to 2008.

Had a bit of a look at exchange (of which I have zero experience):

173GB Mailbox database (See query below).
Circular logging enabled
Logs and database are on the same R5 array (in the same directory a well...)
Backup Exec only runs on the above directory monthly. Windows server backup is not configured and I can't see any other backup software.

We do have mailmarshall which copies all email except internal to internal mailbox and that is backed up daily.

Am I right in saying that with about 20 log files, none older than about 15 minutes on a quiet Saturday - we are in a VERY dangerous position?

The server exchange is running on has 8 bays, all filled, with a single R5 array, and about 250GB free. We have had issues a few years back with running very low on disk space which may be why circular logging has been enabled?

Could I have missed something - is there a built in backup for exchange that may be doing it's bit somewhere? Any suggestions where to look?

Assuming I've not missed something, what's the best plan of attack? I'm thinking:

Backup the database and logs (?) nightly as a separate job.
Switch off circular logging - how does exchange know that a backup has been taken to clear the logs or is that a manual task?
How much disk space should I allow for for the 173GB database?
Shout very loudly at the consultants who haven't mentioned this issue...

Thanks :)
 
Caporegime
Joined
18 Oct 2002
Posts
26,098
VSS-aware backup will handle telling Exchange that it's backed up, so you don't need to worry about that.

175GB isn't a huge amount, could you just chuck it on Exchange Online? If you have to keep it on-prem I would move mailboxes and other roles to a new server before retiring it, RAID 5 makes me uneasy these days.
 
Soldato
OP
Joined
18 Oct 2002
Posts
7,622
Location
SX, unfortunately
We don't appear to have the exchange addition to Backup Exec so not sure what we have currently would work? Minor issue though, it's easy enough to set a reminder to manually delete logs over a certain age...

I'd love to use exchange online of some sort, but they won't pay the cost. There's also over 200GB of PST file archives (yes I know...!).
 
Soldato
OP
Joined
18 Oct 2002
Posts
7,622
Location
SX, unfortunately
They're not going to spring over £10k a year (allowing for some of the mailboxes to be option 2) for something they already own the CALS and exchange licence for. That will come when Exchange 2010 is no longer usable.
 
Caporegime
Joined
18 Oct 2002
Posts
26,098
How many staff do you have running off a single Exchange server if the cost is going to be 10k annually?

Slightly rhetorical question because it's going to be between ~160 and ~340. And you have one Exchange server being managed by you who, no offence intended, appears to be quite new to it all. What's the allowable downtime for when that single box fails?

You should be concerned that you're going to be the one held responsible for outages or loss of data, and the people you report to don't want to throw any budget your way to improve the situation.
 
Last edited:
Soldato
OP
Joined
18 Oct 2002
Posts
7,622
Location
SX, unfortunately
No offence taken, you're absolutely right - but I am only first line, any sign of a meltdown and I'd be on the phone to the external 2nd/3rd line support. I have it in writing that I'm not expected to deal with anything major (define major?). I've just been nosing about to get familiar with it all.

We have 315 mailboxes at present, managers and directors have quite large mailboxes and archives, the rest "need" them but that's really because they can't be bothered to delete stuff they don't need. A whole culture change is needed to sort that. Giving them all 50GB mailboxes wouldn't' help :D

I already have a long term plan to virtualise everything which includes having a second exchange server in a DAG at a different site but I think the above issue needs sorting first ASAP. I need to retire a couple of old server first to free up some licences to make a start on that project. I've been involved in future planning for a while now, it's the day to day I've taken over.
 
Soldato
Joined
27 Feb 2003
Posts
7,173
Location
Shropshire
Personally, circular logging is a last resort when the Exchange logs will fill up the free disk space before the backup can flush them (typically each night). It does mean that in the event of DB corruption, you'll only be able to restore to the last full backup (MS):

In Exchange 2010, circular logging is disabled by default. By enabling it, you reduce drive storage space requirements. However, without a complete set of transaction log files, you can't recover any data more recent than the last full backup. In a normal production environment, circular logging isn't recommended.

If you want to make a simple improvements (and not knowing much else about your backup system), you could get a copy of BackupAssist with it's Exchange Granular add-on. That's under £350 with a upgrades for a year included. Grab a 3 or 5 USB HDs and run a full system image every night to the USB drives which are rotated. If you can spend a bit more and have a suitable location, BA running to a iSCSI target on a Synology or QNAP NAS works lovely as well. Either way, you should then be able to turn off circular logging.

BackupAssist is very easy to setup and with the Exchange option, has some very nice granular restore options : https://www.backupassist.com/backupassist/features/specialized/exchangeserver.html
 
Associate
Joined
7 Oct 2003
Posts
110
Please also have a backup off site, onsite NAS with iscsi you can use Windows backup or backup assist - assist is miles better. Cloudberry for Exchange and then to Amazon S3 very cheap for 200gb your looking at around £5 a month.
 
Soldato
OP
Joined
18 Oct 2002
Posts
7,622
Location
SX, unfortunately
The monthly backup goes onto an iSCSI drive that is connected by fibre to a different building (still not on a removable media though - another thing I need to sort...). I'm using windows server backup to take a copy of the database at the minute. So as to avoid breaking anything I've told it to NOT clear the logs (what little there is). If that works I think I'll make that a daily job until it gets fixed properly.
 
Soldato
Joined
25 Mar 2004
Posts
15,779
Location
Fareham
Is this.. is this running on one server? my lord if so! :D

You can run backup-less if you have multiple copies of the data, see Exchange Native Protection: https://technet.microsoft.com/en-us/library/dd876874(v=exchg.141).aspx#FleMaiPro

Generally this is fine if you meet the following criteria:

You have 3-4 copies of each Database within a DAG. Possibly one of these "lagged" but I think lagged copies makes for annoying admin work.
Preferably not all in the same physical location, i.e. 4 copies with 2 copies in 2 sites.

If the above is true, then enabling Circular Logging is actually recommended. Otherwise it should be disabled and backup jobs run frequently enough to truncate the Logs (these truncation points actually get synced between DAG members, although it usually keeps around 100 logs or so).

The DAG model for this would normally just use JBOD disk assignments for the Database drives, or 1 disk raid groups. If a copy on a server fails you replace the disk, and re-seed it from the other 2-3 working copies. The benefit of this is that you can use cheap SATA or SAS 7.2k RPM disks which don't cost very much.

I would not necessarily virtualise the Mailbox server roles, you'd probably get away with virtualising the HT/CAS roles though. I think a smallish physical presence works best for Mailbox Servers personally, something along the lines of the HP DL380p G8 if your local disk requirements aren't too high, or of course then newer Gen 9's.
 
Soldato
OP
Joined
18 Oct 2002
Posts
7,622
Location
SX, unfortunately
There's no way we will end up with ENP - we're just not big enough. I want a second exchange server for availbility, but that's it. I'm not even going to call it high availability...

Think I'm going to have an interesting discussion with my manager today.
 
Soldato
Joined
25 Mar 2004
Posts
15,779
Location
Fareham
There's no way we will end up with ENP - we're just not big enough. I want a second exchange server for availbility, but that's it. I'm not even going to call it high availability...

Think I'm going to have an interesting discussion with my manager today.

Exchange can be as highly available (or not) as you like, I would suggest the investment should match the importance. If highly available Exchange is important, then some kind of budget should exist for it.

If not save yourself headaches and move it to Hosted Exchange in some form, any respectable provider (include Microsoft) will build a more redundant platform than you can build.
 

Deleted member 138126

D

Deleted member 138126

315 mailboxes running off a single server, with no HA or DR. What is the cost to the business (per hour) of not having mail available? That's where you need to be working from. Even if you had proper backups, how long would it take to get back up and running? A day? Days?

Given your description, 8 bays all full, with (by my calculations) around 500GB of usable space (173GB database + 250GB free)... this sounds like an ancient server. Is it supported? If so, the support alone is probably costing £1000-2000 per year (for any kind of 4 hour response time anyway).
 
Soldato
OP
Joined
18 Oct 2002
Posts
7,622
Location
SX, unfortunately
I can breathe again (a little) - I had a chat with the consultants who pointed out hiding down the bottom of the backup job config is the microsoft store which is backed up every night. So it looks like the most we would lose is a day. Still not good enough in my opinion, but better than a month. Not sure how I missed that, but I'm not familiar with Backup Exec.

The server is a DL380 G7 that is still under 4hr maintenance (think there's just over a year left on the current agreement). The array is 8x300GB in R5 so 2.1 total capacity, split into an OS and Data logical drives. The same server also hosts pretty much everything including file shares etc...

There is a second DC/DNS on an old G5 as well.

Realistically we can afford some down time, it wouldn't kill us to have things down for a day - would cost quite bit of money mind.
 

Deleted member 138126

D

Deleted member 138126

OK, sounds better than initially thought. However, bear in mind that the G7 is not supported for Windows 2012 R2, so you'll have limited upgrade options anyway; i.e. at some point they need to purchase some new hardware, so might as well start thinking what the best long-term options would be.

I'm not a friend of using Windows for file shares, so my preferred option would be Exchange Online (£2.50 per user per month) + a NAS type device. For £10k you could get:

- Low end NetApp
- 2 x DL360 or DL380 to run VMware

- Virtualise the DCs (one per physical node)
- Mail in the cloud
- File shares on the NetApp

Job done

NetApp gives you cloning and replication so covers you for offsite backups.
 
Soldato
OP
Joined
18 Oct 2002
Posts
7,622
Location
SX, unfortunately
The rough plan is:

Use the existing main server in a regional office as a hyper-v host running core services eg. 2nd exchange server, duplicate of file server, DC etc.

Use the DL380p Gen8 we have as a RD server as a host (upgraded to 2 CPUs, additional RAM etc.)
Use the new DL380p Gen9 we have just bought to replace a G5 RD server as a host - the Gen8/Gen9 being the main hosts for the primary VMs.
 
Caporegime
Joined
18 Oct 2002
Posts
26,098
You aren't buying a two node NetApp filer with premium bundle as well as two HP servers for £10k.

If you have a limited budget then I'd be inclined to look at what Server 2016 is doing for Hyper-V and availability that doesn't necessarily require a traditional SAN. If your availability requirements can only be met with shared storage then I'd go for NetApp E-Series connected to the hosts with SAS (no need for iSCSI/FC switches or 10Gb interfaces) and have the backup covered by software.
 
Last edited:

Deleted member 138126

D

Deleted member 138126

You aren't buying a two node NetApp filer with premium bundle as well as two HP servers for £10k.
Sorry, you are absolutely right. I have a quote for the following:

- 1 FAS2520A (cOT 8.3 switchless)
[12x600GB 10krpm] disks
- Licenses : NFS, CIFS, iSCSI, FCP (included), SnapMirror
- Usable Capacity : 3,5 TB (RAID4)
- 2 x 2 ports GbE
- 2 x 2 ports 10GbE

That alone is 10,368.19 €

DL360 Gen8
128 G RAM
4 x 1 GB
2 x 10 GB

Adds 16,492.00 €

Total 26,860.19 €
 
Caporegime
Joined
18 Oct 2002
Posts
26,098
If you could I would be asking where from as that's below our partner promo buy pricing ;)

Edit: In case of confusion I meant the FAS2520 and 2x HP combo for £10k price, not the storage alone. That's still a very keen quotation though.
 
Last edited:
Back
Top Bottom