J. P. Gelb. System managed storage. IBM Systems Journal, 28(1):77--103, 1989.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....file implementation is added, existing files must be checked manually and, where appropriate, moved to the new implementation. Existing programs must be reconfigured before being able to use the new implementation. Gelb examines in detail the problems of exposing file implementations to the user [Gelb89]; note that these problems also apply to VFS, since it too exposes implementations to users. Another alternative approach would be to completely isolate the user from the assignment of files to implementations. Instead, the assignments could be made transparently by the system using a heuristic ....

....implementation or another implementation meant for temporaries. Through parameters, the user provides the extra information needed by the system to make assignments, yet the parameters do not expose low level mechanism to the user. This approach was pioneered in IBM s system managed data product [Gelb89]. Chapter 4 discusses the proposed parameters in detail. 2.5 Summary The file system design sketched in this chapter featured multiple file implementations, allowing it to exploit device parallelism where possible, to utilize idiosyncrasies of different media types, to match the levels of ....

[Article contains additional citation context not shown here]

J. P. Gelb. System managed storage. IBM Systems Journal, 28(1):77--103, 1989. 53

....In many cases, a great deal of additional information is available to describe the goals or behaviors of clients. Such needs can be captured by associating attributes with the storage or its accesses, and used to drive selection of storage device, placement, and dynamic policies such as caching [26, 30, 77]. Research problems here include finding the correct way to specify client needs and storage device behaviors and how best to map one to the other. Service quality specification may be done interactively through an explicit negotiation step, enabling the requesting application to modify its ....

GELB, J.P. System managed storage. IBM Systems Journal 28, 1 (1989), 77-103.

....of device response times for TPCD between a hand generated array configuration and MINERVA generated array configurations (baseline using RAID 5, and RAID 1 0 only) inal configuration in a few cases, the execution times for each of the twelve queries were comparable. 4 Related work Gelb [13] inspired much of our work, by suggesting that the logical view of data be separated from physical device characteristics to simplify the use and administration of storage. To the best of our knowledge, no existing tool will automatically design and configure storage systems. Commercial products ....

J. P. Gelb. System managed storage. IBM Systems Journal, 28(1):77--103, 1989.

....In this approach, referred to as attribute managed storage, the high level application requirements, application workload behavior, and device capabilities are specified by describing attributes of the workload or device. It can be viewed as an extension of system managed storage as defined in [Gelb89]. 1 Workload attributes include performance requirements such as bandwidth, response time, and jitter bounds; resiliency needs such as availability, reliability, and fault models; data sizes, and so on. Some of the attributes specify the performance requirements that must be guaranteed or are ....

J. P. Gelb. System managed storage. IBM Systems Journal, 28(1):77--103, 1989.

....space on the base stations, thus all checkpoints are saved in the stable storage of the home network. 2. 2 Storage Management and Leasing IBM developed a data facility storage management subsystem (DFSMS) that utilized computer technology to reduce the human effort needed to manage storage data [21, 22]. Yin, et al. introduced volume leases for providing server driven cache consistency for large scale distributed systems [23] The leasing approach reduced message traffic at servers for a trace based workload of web accesses. Our approach of adaptive checkpointing with leasing manages storage ....

J.P. Gelb, "System-managed storage", IBM Systems Journal, vol 28, num 1, 1989, pp 77 -- 103.

....space on the base stations, thus all checkpoints are saved in the stable storage of the home network. 2. 2 Storage Management and Leasing IBM developed a data facility storage management subsystem (DFSMS) that utilized computer technology to reduce the human effort needed to manage storage data [21, 22]. Yin, et al. introduced volume leases for providing server driven cache consistency for large scale distributed systems [23] The leasing approach reduced message traffic at servers for a trace based workload of web accesses. Our approach of adaptive checkpointing with leasing makes efficient ....

J.P. Gelb, "System-managed storage", IBM Systems Journal, vol 28, num 1, 1989, pp 77 -- 103.

.... as access latencies) and development (and maintenance) of custom systems for high performance applications [5] Currently existing solutions to large scale storage problems are expensive, both in terms of (a) software costs (HPSS software costs about 100; 000) and (b) people costs (one study [9] reported that on the average, one full time person was needed to manage every 10G bytes of storage ) The problem is getting worse since the costs associated with storage systems are growing (relative to the cost of the raw storage) The build it and then measure it approach to system design ....

J. P. Gelb. System-managed storage. IBM Systems Journal, 28(1):77--103, 1989.

....to needs and minimum overhead allocations . What are appropriate ways to handle the availability implications of client caching . What is the set of mechanisms that should be considered . How can users express their needs in a convenient way 5 Related work IBM s System managed storage model [Gelb89] is in some ways similar. The emphasis of the IBM approach is on allowing a system manager to choose a small suite of mechanisms, and then having the system operate within the constraints these imply (e.g. cylinder and disk placement is left to the system, but the choice of mirroring over parity ....

J. P. Gelb, "System managed storage". IBM Systems Journal 28(1): 77--103, 1989.

....50ms mean time to more bytes and max jitter seconds mean: 50ms jitter: 1ms hints R W ratio fraction of requests that are writes 0.1 request transfer size bytes 256KB access pattern (one of a list) append only Revision 1. 01 23 Brevix generalizes and extends the model of System Managed Storage [Gelb89]. The basis for the design described here was put forward in [Wilkes91] 1 . This has been refined and simplified for use in Brevix. The resilience attributes that can be specified for entities are described in Table 3. Two of the constraints (noncompliance and max data loss) are expressed ....

J. P. Gelb. System managed storage. IBM Systems Journal, 28(1):77--103, 1989.

....and how to balance the load and data across them. Both the workload behavior and the device capabilities are specified by describing attributes of the load or the device respectively. Thus, we refer to our approach as attribute managed storage. We view it as an extension of system managed storage [Gelb89]. Workload attributes include performance requirements such as mean throughput, maximum latency, and jitter; resiliency needs such as availability, reliability, and fault models; cost bounds; data sizes, and so on. Device attributes are expressed similarly. An attribute managed storage system maps ....

J. P. Gelb. System managed storage. IBM Systems Journal 28(1):77--103, 1989.

....In many cases, a great deal of additional information is available to describe the goals or behaviors of clients. Such needs can be captured by associating attributes with the storage or its accesses, and used to drive selection of storage device, placement, and dynamic policies such as caching [26, 30, 77]. Research problems here include finding the correct way to specify client needs and storage device behaviors and how best to map one to the other. Service quality specification may be done interactively through an explicit negotiation step, enabling the requesting application to modify its ....

Gelb, J. P. System managed storage. IBM Systems Journal 28, 1 (1989), 77--103.

....1994; Majumdar 1984; McDonald and Bunt 1989; McNutt 1994; Ruemmler and Wilkes 1991; Ruemmler and Wilkes 1993; Smith 1981] Such a storage hierarchy could be implemented in a number of different ways: Manually, by the system administrator. This is how large mainframes have been run for decades. [Gelb 1989] discusses a slightly refined version of this basic idea. The advantage of this approach is that human intelligence can be brought to bear on the problem, and perhaps knowledge that is not available to the lower levels of the I O and operating 4 . John Wilkes et al. ACM Transactions on Computer ....

....policies for them [Blackwell et al. 1995; McNutt 1994; Mogi and Kitsuregawa 1994] There is a large literature on hierarchical storage systems and the many commercial products in this domain (for example [Chen 1973; Cohen et al. 1989; DEC 1993; Deshpandee and Bunt 1988; Epoch Systems Inc. 1988; Gelb 1989; Henderson and Poston 1989; Katz et al. 1991; Miller 1991; Misra 1981; Sienknecht et al. 1994; Smith 1981] together with much of the proceedings of the IEEE Symposia on Mass Storage Systems) Most of this work has been concerned with wider performance disparities between the levels than exist in ....

GELB, J. P. 1989. System managed storage. In IBM Systems Journal 28, 1. IBM Corp., Armonk, New York, 77--103.

....in practice [Akyurek93, Deshpande88, Floyd89, Geist94, Majumdar84, McDonald89, McNutt94, Ruemmler91, Ruemmler93, Smith81] Such a storage hierarchy could be implemented in a number of different ways: Manually, by the system administrator. This is how large mainframes have been run for decades. [Gelb89] discusses a slightly refined version of this basic idea. The advantage of this approach is that human intelligence can be brought to bear on the problem and perhaps globally better solutions can be developed, using knowledge that is simply not available to the lower levels of the I O and ....

.... on log structured file systems (LFS) Carson92, Ousterhout89, Rosenblum92, Seltzer93, Seltzer95] and cleaning (garbage collection) policies for them [McNutt94, Blackwell95] There is a large literature on hierarchical storage systems and the many commercial products in this domain (for example [Chen73, Cohen89, DEC93, Deshpande88, Epoch88, Gelb89, Henderson89, Katz91, Miller91, Misra81, Sienknecht94, Smith81], together with much of the proceedings of the IEEE Symposia on Mass Storage Systems) Most of this work has been concerned with wider performance disparities between the levels than exist in HP AutoRAID. For example, they often use disk and robotic tertiary storage (tape or magneto optical disk) ....

J. P. Gelb. System managed storage. IBM Systems Journal, 28(1):77--103, 1989.

....In many cases, a great deal of additional information is available to describe the goals or behaviors of clients. Such needs can be captured by associating attributes with the storage or its accesses, and used to drive selection of storage device, placement, and dynamic policies such as caching [26, 30, 77]. Research problems here include finding the correct way to specify client needs and storage device behaviors and how best to map one to the other. Service quality specification may be done interactively through an explicit negotiation step, enabling the requesting application to modify its ....

Gelb, J. P. System managed storage. IBM Systems Journal 28, 1 (1989), 77--103.

No context found.

J. P. Gelb. System managed storage. IBM Systems Journal, 28(1):77--103, 1989.

No context found.

Jack Gelb. System-managed storage. IBM Systems Journal, 28(1):77--103, 1989.

No context found.

Gelb, J.P. System-managed storage. IBM Systems Journal 28, 1 (1989), 77-103.

No context found.

J. P. Gelb. System--managed storage. IBM Systems Journal, 28(1):77--103. IBM, 1989.

No context found.

Gelb, J.P. System-managed storage. IBM Systems Journal 28, 1 (1989), 77-103.

No context found.

J. P. Gelb. System--managed storage. IBM Systems Journal, 28(1):77--103. IBM, 1989.

No context found.

J. P. Gelb. System managed storage. IBM Systems Journal, 28(1):77--103, 1989.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC