SMS: Facts and Fallacies by Mike Kidd Vice President of Marketing, Palindrome Corporation What is SMS? SMS is an open specification designed to solve the major industry problems of managing data in heterogenous environments. It consists of two main components. First, the System Independent Data Format (SIDF), which specifies a standard open tape format as well as a method for "packaging" data independent of the host OS. And second, the Target Service APIs, which along with the standard Target Service Agents (TSAs), give storage management software one consistent interface to all network operating systems. Together, they represent the first workable solution to the proprietary, vendor dependent problems that have plagued the LAN storage management industry. At present, SMS is the only way to reliably back up and restore NetWare 4.01 and NDS (NetWare Directory Service). Why is there so much controversy? To support SMS, a vendor providing backup software must make a fundamental commitment to supporting an open industry standard. All software calls that access a file system directly need to be replaced by standardized SMS calls. The vendor's historically proprietary tape format must be replaced with the standard, extensible SIDF format. As in other parts of the data processing industry, conversion from proprietary solutions to open standards is inconvenient and time consuming to the vendor, but provides huge benefits to the user. When an industry goes through the fundamental change from proprietary to open solutions, there is always a lot of confusion. Any vendor not fully committed to the change will implement "semi-standard" solutions as an interim step, and use any of a number of rationalizations to explain why a fully compliant solution is not available. SMS is at this threshold today. The standard is complete, the tools are in place, but few vendors have yet delivered, creating a fair amount of confusion in the market. Palindrome is now delivering a complete "SMS Level Two" compliant solution, and would like to separate the facts and fallacies surrounding SMS. Fallacy #1: Lots of vendors are fully supporting SMS. Fact: Novell has defined two levels of SMS compliance. Level One describes products that can use standard TSAs to access network clients. Several vendors have shipped products that can use TSAs, although many of them still make direct calls to the file systems, bypassing the standard when convenient. Palindrome's Network Archivist SMS uses only SMS calls, supporting the standard fully. In addition, all TSAs developed and utilized by Palindrome will conform to Novell's specification, meaning that other SMS based applications will be able to interoperate with Palindrome supplied TSAs. Proprietary agents available from other software developers do not interoperate with SMS applications. Some software vendors use a mixture of proprietary agents and SMS TSA's, resulting in increased complexity for users and potential incompatibility with NetWare revisions. Level Two compliance requires that the vendor's software not only use TSAs to access clients, but store their data in the SIDF standard format. Most "SMS based" products on the market today can use TSAs to access data, but lock the data into their proprietary tape formats. To be fair, most have committed to support SIDF at some point in the future, but only as an import/export format - continuing to lock normal backups into the proprietary format. Network Archivist SMS uses SIDF as its native and only tape format, ensuring long-term compatibility with all other SIDF compliant applications. Fallacy #2: SIDF is a moving target. "...SIDF is still a moving target." - Ed Cooper, Vice President of Marketing for Legato Systems, Inc. (LAN Times, 11/1/93) Fact: SIDF is a well-defined, extensible tape format. As new file systems evolve, the format will be extended, but not without backward compatibility. Novell is supporting SIDF with SBackup, and Palindrome, Mountain and System Enhancement Group are supporting it as well. Data interchange on tape is now a reality among those vendors that support the standard. Palindrome is working with the other SIDF compliant vendors to ensure that media written by one can be read by all. Fallacy #3: SMS is a good idea, but it's just too slow. Fact: Palindrome has benchmarked Network Archivist SMS backing up to a DLT tape drive and has achieved sustained throughput of non-compressed data in excess of 100 MB per minute. This is real data (not sparse files) without using NetWare 4.0's data compression. As the industry moves to NetWare 4.x, the on-disk compression will make backups even faster, as SMS based applications can transfer data to tape without decompressing. In these environments, non-SMS applications will actually be much slower. In order to achieve these high speeds, backup vendors may have to endure the inconvenience of changing some of their file access methods. The performance is there for those willing to make the investment. Fallacy #4: SMS won't catch on because it requires backup vendors to change. "It (SMS) imposes Novell's backup philosophy on developers." - Rod Christensen, Vice President of Engineering at Emeritus. (LAN Times, 11/1/93) Fact: Backup as an industry has been among the slowest to adopt standards, but the same forces that have moved other industries to standards-based platforms are at work here. Changing to an open system is never easy or convenient, but it is a growing customer requirement and it is happening. Palindrome has delivered a fully SMS Level Two compliant system, and others will follow in time. (Nearly all have committed to providing Level One support, and most have committed to at least supporting SIDF import/export facilities.) Fallacy #5: There's no way to back up NDS today. "No one is supporting the (NDS) TSA 100 percent, except through emulation mode." - Ed Cooper, Vice President of Marketing for Legato Systems, Inc. (LAN Times, 11/1/93) Fact: Palindrome shipped support for the NDS TSA with Network Archivist 3.0 as a separate utility. With Network Archivist SMS, support for NetWare Directory Services is completely integrated. An entire directory system can be rebuilt via the NDS TSA. Bindery emulation is not a workable solution for backing up NDS. Fallacy #6: SMS isn't "ready for prime time". Fact: SMS has been evolving for two years, led by Novell with significant involvement from a number of backup vendors. The APIs provide a standard, stable and high-performance platform. TSAs available today back up NetWare 3.x (including all name spaces), NetWare 4.01 with NDS, DOS, OS/2 and Windows workstations. In the next few months, TSAs for Macintosh (from Novell), Unix NFS (from Palindrome) and NetWare 2.x (from Novell) will be available. Efforts are underway to provide TSAs for other operating systems and all popular database engines. By the middle of next year, SMS will be the only workable approach to managing storage on complex LANs. Fallacy #7: SMS is just a NetWare-centric specification. Fact: SMS and SIDF are specifications originally developed by Novell to solve the growing problems of protecting data in a heterogenous computing environment. SIDF is now a multi-vendor standard. The SMS APIs are being ported to other environments. Palindrome's upcoming Network Archivist for Unix is based on SMS, using the same APIs and TSA as it's NetWare counterpart. The SMS specification is portable, the first implementation is on NetWare. Summary SMS is here today. Its benefits - standardized access to clients in heterogenous environments, faster support for new releases of NetWare and other operating systems, data interchange and long-term access to data based on an open tape format - are available to any vendor willing to make the commitment necessary to provide full support. If you want to get all the benefits of a fully SMS compliant system, ask your vendor these questions: 1. Can your product use TSAs to access servers and clients? Does it use them exclusively, or does it go around the SMS interface with direct file system calls? A completely standards-based solution will use TSAs, and will adhere to the SMS APIs. Direct access to the file or security system raises the possibility of nonstandard storage of data, and the necessity for a vendor-supplied update when the target operating system changes. 2. Does your product read and write SIDF standard tapes? At the very least, import and export capabilities are required for any hope of data interchange. 3. Does your product use SIDF as its native format? If the answer is "no", then you're locking your backups into a proprietary trap. The only software that will ever be able to restore that data is from your current vendor. Should you change brands, or should they go out of business, your retrieval options are severely limited. 4. Can I do a complete, SMS based backup and restore of a NetWare 4.01 server, including directory services? If not, the SMS application is not complete. The LAN backup industry has remained an island of proprietary formats and protocols. Users have had to wait for each vendor to supply proprietary agents to each new client on the LAN. Backup sessions have been locked into nonstandard formats, limiting user's choices of future backup and storage management applications. Two years ago, SMS and SIDF were launched to provide a way out of these proprietary, incompatible systems. Today, the standard is complete and standards-based backup systems are here. If reliable backup of heterogenous LANs, quick support for new OS releases, data interchange and long-term accessibility of data are important to you, SMS Level Two compliant backup systems are the only answer.