Virtualizing tape library with VTL

VTL is high-speed disk storage presented as tape. It appears to the backup software and application as physical tape library. No changes are required in backup software and backup strategy and process. It eliminates tape sharing between servers, thereby improving restore performance. With no physical tape limitations, more backup jobs can be run in parallel. It avoids mechanical failure so common in tape drives. VTL works in conjunction with tape backup solutions and is not meant to replace it completely.

First generation of VTL suffered from the following drawbacks. Tapes can serve the purpose of remote archival vault, which VTL does not provide that option. Scalability of VTL is very costly as it requires adding controllers or expansion unit. Complexity added to implementation headaches. Inadequate support for data compression due to processing capacity and bus architecture limitations. No support for data de-duplication.

Problem with tape backup are lengthy backup and restore times and media and drive failures. These issues have pushed tape to perform long term archival and DR backup instead of normal production backup.

DataDirect Networks provides a Silicon Storage Appliance (S2A) which is a second generation VTL product. It is particularly suited to large sequential data I/O common in backup operations. For archival or DR purposes, where tape is still required, backup can be first done to this appliance before it can be moved to tapes during non-office hours. The product can also run a backup software (Hosted Backup option), thereby eliminating the need of a backup server and reducing cost. The appliance supports automatic tape caching, where it caches data going to physical tape, increasing performance. It supports data compression before storing it in disks. It is based on commodity SATA storage.

All about disk interface technologies

Serial ATA (SATA)
Used for desktop and laptops. Replacing Parallel ATA because of larger capacity of disks (upto 1 TB), higher speeds (upto 300 MBps with SATA2 technology as compared to 133MBps with PATA) and thinner cables consisting of 7 wires instead of 40 wires for PATA. Thinner cables reduce equipment size and improves airflow, therefore, reducing cooling requirements. SATA drives also require less power than PATA. They are also hot-swappable and cost almost the same as PATA. It uses the transceiver technology used by FC. Speed rates such 300 MBps are burst data rates and not average data rates, which can be 1/2 or 1/3 in some cases of the burst rates. More motherboards are coming with built-in SATA controllers. Introducing SATA in a server requires a SATA controller, SATA drive and a SATA power cable/adaptor. PATA standard is slowly phased out in favor of SATA. CD-ROM/DVD-ROM drives, though, still mostly use PATA interface. 320 GB disk offers the best price per GB for now. Drive cache is another consideration for purchase. Drives now have 8 - 16 MB caches. SATA cables can be upto 1 meter long as compared to 18 inches for PATA cables. SATA drives are of the highest capacity as compared with any disk technology. Mostly used for backup and archiving and other non-mission critical (enterprise secondary storage) purposes. Reduced crosstalk as compared to PATA because of lower density of wires. They are lowest in cost as compared to SAS and FC devices. Serve in half-duplex mode.

Parallel SCSI
A popular and long lasting disk technology is now reaching its end of production use. Ultra320 SCSI is the last progression offering speeds of 320 MBps. Its access time of close to 5.2-5.7 msec is better than 9-12 msec of PATA drives. Supports upto 15 devices per channel/bus. Serve in half-duplex mode. Each device identification is set after installation and must be unique on the same bus. Proper termination is a requirement. Cables have 68 pins. Mostly used as internal storage of servers. It is generally not hot-pluggable. Parallel SCSI suffers from cabling, voltages and connector size challenges.

Serial SCSI (SAS)
SAS point-to-point technology is rapidly replacing parallel SCSI at entry level servers. It is also a replacement of FC storage on low to mid-range storage devices. Current speed is around 300 MBps (3 Gbps) and it has a roadmap for upto 1200 MBps speeds. Single cable support four channels. Serve in duplex mode. Each device id is set in the factory and is globally unique. Can support 128 devices and upto 16,000 devices with expanders (SAS switches). They are available in compact 2.5 inch form factor. Cables have only 7 pins. SAS enclosures support both SATA and SAS drives as it supports SATA interconnect technology. SAS and SATA drives together in the same enclosure can fulfill tiered storage requirements, where faster SAS drives can be primary storage and cheaper SATA can store older data. Used as both internal and external directly attached storage (DAS). It is hot-swappable. SAS supports three protocols, SMP for managing point-to-point topology, SSP (serial SCSI protocol) allows controller to communicate with SAS devices using SCSI software and commands and STP (serial tunneling protocol) allows SAS controllers to communicate with SATA devices. Upgrade from parallel SCSI to SAS protects existing investment in SCSI software. It provides support for dual-ported devices. Designed for mission critical use and have a bright future ahead.

Fibre Channel (FC)
Mostly deployed in mid-range and high-end environments. As of now, considered the fastest technology with speed upto 400 MBps (4 Gbps). Only FC enclosures can be used in a SAN environment. These enclosures can also support FC and SAS drives. FC drives use SCSI command set.