Lync 2013 Sizing – The HA Discussion

When going through the sizing guide for Lync 2013 I came across an interesting aspect of how many Users I should consider to load based on the model of HA I Choose….The Possibilities became very Interesting…My First Attention went to the FE Server sizing and focus shifted from sizing to the different models of HA possible and Logic to decide this….

For this article let’s denote ‘N’ as the number of Servers required for a particular requirement.

I’ve always used the notation of N+1 and N+N but found that different people have different definitions for these.

So Lets first zero in on what I mean in my version of these two models:

N+1

In This model we only provision for 1 additional server than the required number. What Happens here is that when one server fails all servers work at full capacity. If any other server fails then the collapse of the entire pool is imminent unless the solution has some sort of mechanism to lower performance but retain operations.

N+N

In this model we have equal number of server as required. The key advantage of this model is that it can tolerate upto 50% server collapse. The Negative is that the Pool Capacity is halved and Costs of Compute doubles.

Now let us look at the different maximum capacities we should consider for FE Sizing.

In Lync Server 2013 Each FE Pool can have a maximum of 12 Servers and each Server can take a load of maximum 6660 Simultaneous Users. Now when I try to split this across different overall Loads that I want to handle, the results looks interesting…

Per Pool Scale

HA Option

 

N+1 Per Server Load on Failure

 

N+N Per Server Load on Failure

Total Users

No HA

N+1

N+N

 

1

2

3

4

5

6

 

1

2

3

4

5

6

6660

1

2

2

 

6660

 

 

 

 

 

 

6660

 

 

 

 

 

13320

2

3

4

 

6660

13320

 

 

 

 

 

4440

6660

13320

 

 

 

19980

3

4

6

 

6660

9990

19980

 

 

 

 

3996

4995

6660

9990

19980

 

26640

4

5

8

 

6660

8880

13320

26640

 

 

 

3806

4440

5328

6660

8880

13320

33300

5

6

10

 

6660

8325

11100

16650

33300

 

 

3700

4163

4757

5550

6660

8325

39960

6

7

12

 

6660

7992

9990

13320

19980

39960

 

3633

3996

4440

4995

5709

6660

46620

7

8

12

 

6660

7770

9324

11655

15540

23310

 

4238

4662

5180

5828

6660

7770

53280

8

9

12

 

6660

7611

8880

10656

13320

17760

 

4844

5328

5920

6660

7611

8880

59940

9

10

12

 

6660

7493

8563

9990

11988

14985

 

5449

5994

6660

7493

8563

9990

66600

10

11

12

 

6660

7400

8325

9514

11100

13320

 

6055

6660

7400

8325

9514

11100

73260

11

12

12

 

6660

7326

8140

9158

10466

12210

 

6660

7326

8140

9158

10466

12210

79920

12

12

12

 

7265

7992

8880

9990

11417

13320

 

7265

7992

8880

9990

11417

13320

What you can notice is that the N+1 configuration is capable of handling only single server fail. However using N+N is not blind solution as well. The Max Failover capability peaks when the need is for 33,300 to 46,620 users total which equates to a capability of handling 5-6 server fails.

Now my question at this point is if this should be ideal size (33K to 46K) per pool, instead of the maximum 80,000…what do you think?

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

w

Connecting to %s