Start a Conversation

This post is more than 5 years old

Solved!

Go to Solution

9982

November 25th, 2010 03:00

E20 -001 questions

hi guys

i was wondering if someone could tell me how you work out these questions! id really appreciate help on these as im fine with the rest of the book, just cant get these!!

1. An application uses ten 9 GB devices. A full volume local replica of the app is required every 6 hours. Each replica will be kept for 24 hours, how many replica devices are required?

2.  A disk has cylinder Head Sector (CHS) valued of 1000, 4 and 25. What address range will be used for the logical block address (LBA) Values for this disk?

3. You have five disks in a raid 0 set and have assigned a striped depth of 128 blocks (64 kb) . What is the stripe size (KB)?

4 . An application used ten 9 GB devices. A pointer based replica of the application is required every 4 hours. Each replica will be kept for 24 hours. The data changes by 10% every 4 hours. How many replica devices are required ?

please??

December 2nd, 2010 11:00

Hi,

I saw my name. Regarding Question 3.

In ISM terminology we have the following equation:

{stripe_depth} x {number_of_drives_in_the_RAID_set} = {stripe_size}

so:

64KB x 5 = 320KB

Note also that, in ISM terminology:

  • stripe_depth = strip size (no 'e' in strip)
  • number_of_drives_in_the_RAID_set = stripe_width

So, an equivalent equation is:

{strip_size} x {stripe_width} =  {stripe_size}

So, just be careful and realise that one little 'e' character makes all the difference: strip_size is not the same thing as stripe_size.

It might help also if you think about the units - (integer) x (KBs) is always (KBs).

To be a little picky (sorry), there is no ISM term called 'striped depth'.

Does this make sense ?

Thanks, Richard.


2 Intern

 • 

546 Posts

November 29th, 2010 09:00

Thanks Mike for editing the questions a bit --- I'll take a swipe at a couple of them, and let the troops come in and help with the rest of them.

For me, drawing things out helps alot. So here's the picture I drew for question 1:

question1.gif

So first take all the extra stuff out of the question. The question is "how many replica devices are required". So its asking how many devices, not size, so I don't care about the 9GB part. I know the application uses 10 devices. I need a full replica every 6 hours, or 4 times a day. I need to keep each replica for 24 hours. So to keep 4 full replicas -- which is 10 disks per replica -- I need 40 devices.

For question 2, check page 32 - if someone wants to take a crack at that picture I'd appreciate it!!

For question 3, check page 55 in the ISM book. (I think this is Richard Butler's favorite topic -- so I'll ping him!!)

For question 4 -- you want to take a crack at making this picture?

December 2nd, 2010 12:00

Hi mikeleahy,

Regarding Question 2, have a look at the following from a post I made last year:

Each physical disk block has a unique physical  address - CHS. (Actually, in this scheme, CH identifies a unique track,  and so CHS identifies a unique disk sector within that unique track).  For an open systems disk, the sector size will be 512B.

When the  drive is formatted, each disk sector (CHS) is assigned a logical block  address. Also, when the drive is formatted, then it is formatted 'down' a  cylinder, so logical block 0 -> logical block N-1 all exist within  one cylinder (where N is the number of sectors in that cylinder).  Logical block N is then located at the start of the next cylinder, and  this next cylinder contains disk sectors N -> 2N-1.

(BTW, the  drive is formatted 'down' a cylinder in order to improve the sequential  throughput (MB/s) of the drive; the drive reads the sectors in the first  cylinder, and this is purely an electro-magnetic operation. We then  have the electro-mechanical operation of the moving the heads to the  next track, and then we can read the next cylinder. In contrast, the  random I/O performance of the disk (IOPS) is improved through the use of  the command tag queuing technology).

This post might help, let us know (I did not want to post the exact answer, because if you work through this, and draw a diagram, I think you will learn a lot more and it will stay in your memory better).

Also of interest is the following post, which aludes to the direction in which the cylinders are formatted.

https://community.emc.com/message/13460#13460

Thanks, Richard.

2 Intern

 • 

546 Posts

December 6th, 2010 07:00

Hi - did these responses help answer your questions? thx -Gina

1 Message

February 5th, 2013 07:00

Hi,

There is an example in prectice test of EMC ISM v2:

"An application uses ten 15GB devices. A pointer based full volume replica of the application is required. The replica will be kept for 24 hours and the data changes by 10% every 24 hours. How much storage should be allocated for the replication?"

Answer : 150 GB.

As per above calculations 15GB should be the answer.

How it is calculated?

17 Posts

May 23rd, 2013 03:00

Hey Nandarshi,

When I came across this question, I thought the same as you. It's quite logical that since you have only one replica, 150 GB of total storage and 10 % of changes, 15 GB of actual space is required for a replica.

However the question sounds quite tricky: "How much storage should be allocated for the replication?"

A pointer based full volume replica requires storage which is at least the same size as the source, even if 90 % of it would not be used.

No Events found!

Top