How to Know if Virtual Tape Technology is Right for You Chris Dedham Business Continuity Specialist and Senior Storage Solutions Architect Mainline Information Systems

Review of Backup Topologies - Backup directly to Tape Database Tape Tape Duplication for DR The disk storage contains the production data Host Data is backed up directly to tape Tape Disk

Review of Backup Topologies - Backup directly to disk, then moved to tape [D2d2T] Database Tape Tape Duplication for DR The disk storage contains the production data Host Data is backed up directly to disk Tape Data is flushed from disk to tape Disk Disk

Review of Backup Topologies - Backup directly to disk, and stored on disk [D2D2T] Database Tape Copy for DR and Archiving The disk storage contains the production data Host Data is backed up directly to disk and stored on disk Tape Disk Disk or VTL

Review of Backup Topologies - Backup directly to VTL, and replicated to VTL Database The disk storage contains the production data Host Disk VTL VTL Data is backed up directly to Disk on a VTL, and stored on the VTL De-Duplicate the data and replicate it to another VTL over the WAN “ Truly Tape Less”

Review of Backup Topologies - Backup directly to VTL, and replicated to VTL Database The disk storage contains the production data Host Disk VTL VTL Data is backed up directly to Disk on a VTL, and stored on the VTL De-Duplicate the data and replicate it to another VTL over the WAN “ Truly Tape Less” Tape for Archiving Tape

Why are customers interested in VTL’s A desire to get away from Real Tape Tape drive and tape automation break - fix issues Eliminate tape handling – Operational / Security Doing something different than status quo Increase backup throughput More tape drives (aka mount points) for greater parallelism LAN Free to disk Helps with small files and NDMP Less administrative overhead than large diskpools

A desire to get away from Real Tape

Tape drive and tape automation break - fix issues

Eliminate tape handling – Operational / Security

Doing something different than status quo

Increase backup throughput

More tape drives (aka mount points) for greater parallelism

LAN Free to disk

Helps with small files and NDMP

Less administrative overhead than large diskpools

Why are customers interested in VTL’s (cont) Faster restores No tape mount load times (Disk seeks instead) Tape volume fragmentation is greatly reduced Increase TSM admin throughput processing Reduced disk pool migrations Faster DR copies to real tape Faster reclamation processing Additional drives (aka mount points)

Faster restores

No tape mount load times (Disk seeks instead)

Tape volume fragmentation is greatly reduced

Increase TSM admin throughput processing

Reduced disk pool migrations

Faster DR copies to real tape

Faster reclamation processing

Additional drives (aka mount points)

Why are customers interested in VTL’s (cont) Replication for Disaster Recovery Tape handling is costly operationaly and can create security issues Data De-Duplication can greatly reduce the amount of data that must be transferred over the WAN WAN costs can be 50% to 70% TCO of a DR solution Data De-Duplication can reduce the cost of disk Making 30TB’s of disk look like 300TB’s

Replication for Disaster Recovery

Tape handling is costly operationaly and can create security issues

Data De-Duplication can greatly reduce the amount of data that must be transferred over the WAN

Data De-Duplication can reduce the cost of disk

Making 30TB’s of disk look like 300TB’s

Some facts about VTL’s , however Real tape can be faster than virtual tape High speed tape can be faster than virtual tape depending on the workload Real tape can store data at a lower cost per GB Storing inactive data on real tape can have better environmental characteristics Virtual tape may require “straight disk” and real tape to handle all of the workload

Real tape can be faster than virtual tape

High speed tape can be faster than virtual tape depending on the workload

Real tape can store data at a lower cost per GB

Storing inactive data on real tape can have better environmental characteristics

Virtual tape may require “straight disk” and real tape to handle all of the workload

The special sauce in VTL’s is intelligent compression (aka Data De-Duplication) All VTL’s can do standard LZ compression like in tape drives LZ compression reduces space within files. Typically 2:1 compression ratio Some VTL’s have Data De-Duplication Data De-Dupe eliminates redundant files 7:1 or greater De-Duplication ratios can be obtained “ Your mileage may vary”

All VTL’s can do standard LZ compression like in tape drives

LZ compression reduces space within files.

Typically 2:1 compression ratio

Some VTL’s have Data De-Duplication

Data De-Dupe eliminates redundant files

7:1 or greater De-Duplication ratios can be obtained

“ Your mileage may vary”

There are two methods for Data De-Duplication In-Line Data De-Duplication The incoming backup data is de-duped before it lands on the VTL storage. The IBM TS7650g (Diligent) and Data Domain use this method Post Processing Data De-Duplication The incoming backup data is de-duped after it lands on the VTL storage The FalconStor VTL and Sepaton VTL use this method

In-Line Data De-Duplication

The incoming backup data is de-duped before it lands on the VTL storage.

The IBM TS7650g (Diligent) and Data Domain use this method

Post Processing Data De-Duplication

The incoming backup data is de-duped after it lands on the VTL storage

The FalconStor VTL and Sepaton VTL use this method

Three Basic Approaches Talked about today in the industry: Hash based de-duplication Sometimes referred to as a Content Addressable Storage approach Content Aware Assumes the best candidate to de-dupe against is an object with the same properties (name etc.) HyperFactor A different approach based on an agnostic view of data

Talked about today in the industry:

Hash based de-duplication

Sometimes referred to as a Content Addressable Storage approach

Content Aware

Assumes the best candidate to de-dupe against is an object with the same properties (name etc.)

HyperFactor

A different approach based on an agnostic view of data

How Hash Data De-duplication (DDD) Works 1. Data chunks are evaluated to determine a unique signature for each 2. Signature values are compared to identify all duplicates 3. Duplicate data chunks are replaced with pointers to a single stored chunk, saving storage space C A B C A A B B A Data Store Data Store C A B C A A B B A Data Store C A b c a b B a a Data Store

Identification of Redundant Chunks Unique identifier is determined for each chunk Identifiers are typically calculated using a hash function that outputs a digest based on the data in each chunk MD5 SHA For each chunk, the identifier is compared against an index of identifiers to determine whether that chunk is already in the data store Selection of hash function involves tradeoffs between Processing time to compute hash values Index space required to store hash values Risk of false matches

Unique identifier is determined for each chunk

Identifiers are typically calculated using a hash function that outputs a digest based on the data in each chunk

MD5

SHA

For each chunk, the identifier is compared against an index of identifiers to determine whether that chunk is already in the data store

Selection of hash function involves tradeoffs between

Processing time to compute hash values

Index space required to store hash values

Risk of false matches

Deduplication Ratios Used to indicate compression achieved by deduplication If deduplication reduces 500 TB of data to 100 TB, ratio is 5:1 Deduplication vendors claim ratios in the range 20:1 to 400:1 Ratios reflect design tradeoffs involving performance and compression Actual compression ratios will be highly dependent on other variables Data from each source: redundancy, change rate, retention Number of data sources and redundancy of data among those sources Backup methodology: incremental forever, full+incremental, full+differential Whether data encryption occurs prior to deduplication

Used to indicate compression achieved by deduplication

If deduplication reduces 500 TB of data to 100 TB, ratio is 5:1

Deduplication vendors claim ratios in the range 20:1 to 400:1

Ratios reflect design tradeoffs involving performance and compression

Actual compression ratios will be highly dependent on other variables

Data from each source: redundancy, change rate, retention

Number of data sources and redundancy of data among those sources

Backup methodology: incremental forever, full+incremental, full+differential

Whether data encryption occurs prior to deduplication

And data deduplication is the key to using more disk more cost effectively!

Thank you Knowledge is POWER at: Mainline’s Knowledge Center www.mainline.com/kc 866.490.MAIN (6246)