Exchange 2016 design/resources (upgrading from 2010)

Hi experts.

We're upgrading from Exchange 2010 to 2016 and I already have the design on my head, which will be similar to what we have on 2010, but I'm unsure regarding how many servers will we need to accommodate our mailboxes and run smoothly.

I've tried to use the  Exchange Server Role Requirements Calculator v9.1, but I'm not sure why it keeps giving me so many servers that it can't be right!
The idea is to have enough servers to run smoothly even in the case of 1 of them failing (DAG)
If someone could give me some thoughts on the below, that would be much appreciated.

Six databases, with around 200GB each
Total number of users is 1800, around 300 mailboxes per database
Average size of each mailbox is around 400mb (limit is 3GB) (they're small because we have a third-party archiving solution stubbing messages)
Growth: I don't expect us to grow a lot in mailbox numbers, let's say 25% in the next 2 years. In terms of mailbox size, I would say each mailbox could grow up to 50% in the next 2 years, so, each database would go up to 300GB or 350GB.
Outlook - 80% users will work in Online Mode and 20% in Offline Mode.
Daily email flow - Average of 5000 outbound and 12000 inbound emails. In terms of size, they're mostly small emails with a few kb.
Servers will be virtual machines in VMware.
Hosts servers have CPU Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz (logical processors 32) and no RAM restrictions (can go up to 92GB).

One Active/Passive DAG with 2 servers in production (active) and 3 servers on a failover site (passive).
The 2 servers in PROD will have 1 copy of all databases and on a day-to-day operations will run 3 active and 3 copies with preference 2, so they get activated if the other server crashes (basically, load will be balanced between 2 servers).
On the failover site we'll have 2 servers with 1 copy of all databases and preference 3 and 4 and activation blocked, so they only get activated manually (in case the prod site dies). The third server on the failover site will have lagged database copies and will only be used in case database corruption happens on the other servers and we cannot recover them.

So, my 3 main questions are:
1. Will 2 servers in production be enough to run these 6 databases smoothly (3 databases in each server)?
2. Will 1 server be enough to run all 6 databases smoothly (in case the other one crashes)?
3. Do you recommend a different design/solution?

Please let me know if you need more info. I can always draw the design and attach it, to help visualising this scenario.

Thank you in advance.
Adriano GasparAsked:
Who is Participating?
MaheshConnect With a Mentor ArchitectCommented:
Attached here is revised calculator sheet and Specint2006 values for your processor
I noticed few things:
you have kept minimum database size 500 MB and maximum 3GB, it look like 6 times
better if you could simply done tiring as few users would need big mailboxes and other need standard size say 500 to 1 Gb and few 500 Mb to 3 GB, it will save storage
Also as far as I know, no need to keep lagged copies, you need to ensure that backups are happening as appropriate as if database level corruption occurs , it cannot be noticed immediately and in that case even lagged copy will also have it, in that case hard database repair is the only option or you need to hit Microsoft door, Lagging database alone need two servers in DR only to store standby copies
Also I have made total 3 total copies per database, 2 in primary DC and 1 in DR site
if you removed lagged copy, you need only 6 servers across DAG, you need to enable cross site failover so that primary data center can sustain failure up to 3 nodes and you will get enough time to build new servers and add them to DAG, otherwise DAG can sustain only 1 node failure post which databases will start dismounting as cross site failover is disabled
This is as per my view, you need to decide

if mailbox limit is 3 Gb / mailbox then you need 1800 × 3 gb = 5400 gb of total storage
plus you need to add storage and resources (CPU and memory) based on % growth expected in users
you need minimum 2 servers in production with active and passive copies
so, to arrive correct design you need to enter above along with send / receive messages per day in calculator and then it will give you number of servers required at DR
Adriano GasparAuthor Commented:
Thanks for your comment, Mahesh.
I've tried using the calculator several times and it seems straightforward to add the values, but or it says I need something like 16 servers or gives me error messages.
At the moment it's giving me the error "The design parameters you have chosen have resulted in more mailbox servers being required to support the design than what a DAG can support".
I cannot figure out what is wrong!
Another thing I cannot find is our CPU in the website! So, I don't know what the SPECInt2006 rate is! I've put an average value.

I've updated the calculator file to this comment, do you want to have a look?
Thank you.
Has Powershell sent you back into the Stone Age?

If managing Active Directory using Windows Powershell® is making you feel like you stepped back in time, you are not alone.  For nearly 20 years, AD admins around the world have used one tool for day-to-day AD management: Hyena. Discover why.

send me CPU spec and PROFILING data as well
also send send / receive rate
Adriano GasparAuthor Commented:
Hi Mahesh, I hope this is all the necessary info. I've attached the calculator xlsx file on my last comment, but it seems it didn't attach.
I'll attach it again. Anyway, below is the users tier and CPU details.

The CPU is an Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz (logical processors 32). I think they have 4 cores

Total Number of Tier-1 User Mailboxes / Environment = 1800
Projected Mailbox Number Growth Percentage = 25%
Total Send/Receive Capability / Mailbox / Day = 50 messages
Average Message Size (KB) = 75kb
Initial Mailbox Size (MB) = 500MB
Mailbox Size Limit (MB) = 3072MB
Personal Archive Mailbox Size Limit (MB) = NO Archive
Deleted Item Retention Window (Days) = 14
Single Item Recovery = Yes
Calendar Version Storage = Yes
Multiplication Factor User Percentage = 30% (maybe less)
IOPS Multiplication Factor = 1.00
Megacycles Multiplication Factor = 1.00
Desktop Search Engines Enabled (for Online Mode Clients) = Yes
Predict IOPS Value = Yes
I will get back to you shortly
Adriano GasparAuthor Commented:
Hi Mahesh.
Thank you for the Specint2006, much appreciated. Now with this value, the calculations seem to be ok!

I think I'm going to follow your advice and not have lagged copies. We have daily backups, and the corruption is not immediately noticed, so, I would have to have lagged copies several hours behind and in that case, we can use the backups.

I'm also going to create different user profiles, so, I can have less storage. For this, I'll get a list of all mailboxes and their current sizes.

Regarding the DAG and number of servers, the goal of having a secondary site/datacenter is to only use it in case of complete failure on the primary site, therefore, we want to mirror what we have on the primary site.
That said, I prefer to have 3 servers on the primary site and the exact same on the secondary site. This will mean that, having 6 databases, we would have 2 active databases and 4 passive databases on each server on the primary site (and also on the secondary site).

Thank you for your comments Mahesh and for the CPU Specint2006, appreciated.
MaheshConnect With a Mentor ArchitectCommented:
The purpose of DR setup is to temporary survival / arrangement until primary data center come up
its not 100% successful as rest of the environment is not fully ready / prepared while disaster happens and as a thumb rule, the server count at DR should be 50% of primary data center, it should not be equal.
If you reduce server count to 3 at production, DR you should have 2 servers as per calculator, 3 servers are not required
Adriano GasparAuthor Commented:
HI Mahesh and thank you for your comments.
I'll pin you as the best solution. Thanks
You are welcome
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.