17.03.2015 Views

Storage Area Networks For Dummies®

Storage Area Networks For Dummies®

Storage Area Networks For Dummies®

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

350<br />

Par t V: Understanding the Cool Stuff<br />

The trade-off is the size of the index table and the number of entries required.<br />

As you see in Figure 13-11, fewer hashes are required when you use larger<br />

blocks; therefore, the index table will be smaller and faster. On the other<br />

hand, your de-dupe ratio is not as efficient when you use larger blocks.<br />

Always ask your vendor whether its solution is tunable based on your policies,<br />

so that you can determine whether you want to use more disk space<br />

and fewer hash entries to get a faster solution, or use less disk space and<br />

more hash entries to get a solution that is a bit slower but generates much<br />

better de-dupe ratios. Creating a policy is especially useful for solutions that<br />

work out of band; because performance is less of an issue (the backup job is<br />

already complete) you can use more time but get a better de-dupe ratio by<br />

using smaller chunks.<br />

Hash Updates to Index Table<br />

Using 4KB Chunks = 3 Entries<br />

Hash Entries in Index Table<br />

Using 512-byte Chunks = 24 Entries<br />

Figure 13-11:<br />

Hash index<br />

table entries<br />

using<br />

different<br />

data chunk<br />

sizes.<br />

74476bea33bce9985dbf96<br />

23476bea872cd9985bcad3<br />

9757fa23472cd9985b76bc<br />

Hashes stored in the index<br />

9757fa23472cd9987c98bc<br />

fb5761298ca6597367234a<br />

5e7623ac6981bd4d32a972<br />

bce6df3987a234a78923c6<br />

7664baf4987a675cbff3269<br />

75bcd4eaf39872fccb46aa<br />

897cde458712accff76499<br />

7664baf4987a675cbff3269<br />

75bcd4eaf39872fccb46aa<br />

897cde458712accff76499<br />

75bcd4eaf39872fccb46aa<br />

9757fa23472cd9987c98bc<br />

fb5761298ca6597367234a<br />

5e7623ac6981bd4d32a972<br />

5e7623ac6981bd4d32a972<br />

7664baf4987a675cbff3269<br />

75bcd4eaf39872fccb46aa<br />

897cde458712accff76499<br />

7664baf4987a675cbff3269<br />

7779b2c3598cadef319876<br />

7779b2c3598cadef319876<br />

7779b2c3598cadef319876<br />

9757fa23472cd9987c98bc<br />

7779b2c3598cadef319876<br />

7779b2c3598cadef319876<br />

7779b2c3598cadef319876<br />

Hashes stored in the index<br />

What about hash collisions?<br />

A hash collision isn’t a traffic accident, but it could have the same ramifications<br />

for your business. A hash collision happens when the resulting hashes for<br />

two unique chunks of data are the same. Data de-duplication finds redundancy<br />

by hashing blocks of data and comparing the hashes to determine whether<br />

two data blocks are the same or unique. In this example, two similar inputs

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!