This post is more than 5 years old
19 Posts
1
3383
E20-324 FAST question
would like to know how to go about doing this calculations for the following question thanks
customer has a heterogenous pool with the following config:
1000GiB of flash, 50 Gib free
5000Gib of SAS, 1000 Gib free
50000Gib of NL-SAS, 30000 Gib free
the customer creates a new lun with data placement policy set to "auto-tiering". The LUN is then populated with 400Gib of data. How is data distributed?
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
2
April 11th, 2013 12:00
Yep, starting in Flare 32 it starts at the highest available tier then auto tiers.
You can read more about it on page 16 of the FAST VP white paper:
https://support.emc.com/docu32691_White-Paper:-EMC-FAST-VP.pdf
So to answer your inital question, 50 GB of the 400 GB would land on SSD and the remainder would land on your middle tier.
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
March 12th, 2013 07:00
In Flare 32, when creating a lun with auto-tiering enabled, the data will be located on the highest available tier. In this case, 50 GB would be on flash, and 350 GB would be on SAS drives.
Karthik_Kallam
4 Posts
0
April 11th, 2013 10:00
If "auto-tiering" is enabled, data is distributed evenly. In this example 50:1000:30000=> 1:20:600
So 400GB is distributed as 0.64GB in Flash, 12.8 GB in SAS and 386.56GB in NL_SAS
Calculation:
1x+20x+600x=400GB
621x=400GB
x=400/621GB
multiply x with respected ratios
1x:20x:600x => 0.64 : 12.8 : 386.56(Approx)
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
2
April 11th, 2013 12:00
Karthik,
It's my understanding that starting in Flare 32 it no longer distributes the data evenly as it did in previous verisons of Flare code. Would you concur?
vforde
67 Posts
1
April 11th, 2013 12:00
You are right as I understand too. The new recommended policy is Start High then Auto-Tier. See this video 09:10 confirming same VNX Video - FAST VP - YouTube i.e. flash first
Regards,
Victor
RRR
2 Intern
2 Intern
•
5.7K Posts
1
April 12th, 2013 02:00
flash first! You are correct.
pokus123
9 Posts
0
July 2nd, 2013 10:00
There were problems in few environments, if I remember well, customers who had lots of writes had a problem if tiering was enabled, because of using lowest available tier prior code 30? Not sure... And thats not a good idea using NL-SAS from pool for huge random write peaks... So now should be as it is described here - highest available tier...
golnaz1
16 Posts
0
August 12th, 2013 21:00
Hi ,
I have a similar question in e20-324 , I got confused as the answer is different from what I think is correct, here is the question:
customer has a heterogenous pool with the following configuration:
1000 GiB of flash, 400 Gib free
10,000 Gib of SAS, 2,000 Gib free
80,000 Gib of NL-SAS, 20,000 Gib free
the customer creates a new lun with data placement policy set to "auto-tiering". The LUN is then populated with 500Gib of data. How is data distributed?
a. 300 GiB on Flash , 200 GiB on SAS , no data on NL-SAS
b. 166 GiB on Flash , 166 GiB on SAS , 167 data on NL-SAS
c. 9GiB on Flash , 100 GiB on SAS , 446 data on NL-SAS
d. 400 GiB on Flash , 100 GiB on SAS , no data on NL-SAS
The correct answer is c . But what i think is correct is the highest available first ..
Please help me for the correct answer.
Thanks,
christopher_ime
2K Posts
2
August 17th, 2013 19:00
The answer would be "c" except that you had copied the answer down incorrectly for the SAS tier. The values you have add up to 555 GiB
NOTE: Technically there is roughly 2% overhead for pool LUNs but they aren't testing you on this.
The answer would be (rounded up):
9 GiB Flash
45 GiB SAS
446 GiB NL-SAS
The reason it would not be "d" is because even in VNX OE for Block v32 there are 2 separate policies:
1) Start High then Auto-Tier (recommended)
2) Auto-Tier
The question states: "Auto-Tier" only.
The math would be as follows:
1) Ratio of free space = 400:2000:20000
2) Reduce to: 1:5:50
3) Calculate:
a)1x + 5x + 50x = 500
b) 56x = 500
c) x = 8.92
4) Calculate each tier:
a) Flash = 1x = 9 GiB
b) SAS = 5x = 45 GiB
c) NL-SAS = 50x = 446 GiB
golnaz1
16 Posts
1
August 17th, 2013 19:00
Thank you Christopher Imes for your complete answer .
christopher_ime
2K Posts
1
August 17th, 2013 23:00
No problem. Unfortunately though I'm struggling to calculate the other question you posted.
golnaz1
16 Posts
1
August 21st, 2013 00:00
Hi Christopher Imes,
so here is the question for "highest available" . Appreciate if you can help with that.
A customer has a heterogeneous pool with the
following configuration:
-800 GiB of Flash, with 300 GiB free
-12,000 GiB of SAS, with 2,000 GiB free
-50,000 GiB of NL-SAS, with 25,000 GiB free
The customer creates a new LUN with data placement policy set to "highest available tier". The LUN is
then populated with 600 GiB of data.
How is that data distributed (numbers are approximate)?
A. 220 GiB on Flash, 380 GiB on SAS, no data on NL-SAS
B. 300 GiB on Flash, 400 GiB on SAS, no data on NL-SAS
C. 200 GiB on Flash, 200 GiB on SAS, 200 GiB on NL-SAS
D. 10 GiB on Flash, 110 GiB on SAS, 480 GiB on NL-SAS
christopher_ime
2K Posts
1
August 24th, 2013 23:00
The answer is A.
The question is testing you on whether or not you know that the system will (try to) maintain 10% free per tier.
The calculations will be:
1) 800 GiB * .10 = 80 GiB (10% of Flash will be kept free in this example)
2) 300 GiB - 80 GiB = 220 GiB on Flash
3) 600 GiB - 220 GiB = 380 GiB on SAS
golnaz1
16 Posts
0
September 19th, 2013 22:00
Thank you for your help!