ANDERSON,D.,CHASE,J.,AND VAHDAT,A. 2000. Interposed request routing for scalable network storage. In Proceedings of the Symposium on Operating Systems Design and Implementation.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....and work at the edges of the network because of the processing overhead involved. There is also the overhead of extra messaging with the external cache engines [36] SEN supports caching for different data types by identifying all types as generic objects. Slice architecture by Anderson et al. [13] provides network file service in LANs with network attached storage. proxy is a component in Slice L5 protocol that provides content based request switching. proxy is implemented as an IP packet filter and can reside within the networks , but it must still reside (logically) at the end of a ....

D. C. Anderson, J. S. Chase, and A. M. Vahdat. Interposed request routing for scalable network storage. In Proceedings of the 4th Symposium on Operating Systems Design and Implementation (OSDI), Oct. 2000.

....connection across multiple IP paths. Thus the multi path QoS routing results found in the literature can be used in conjunction with a connection striping scheme. Bandwidth aggregation has also been considered in the context of storage systems and high volume data or file servers, for instance [17, 18, 19, 20]. Typical storage server architectures are confined to within a LAN having a single controlling authority. Note that within a LAN it is possible to do bandwidth aggregation at the MAC layer. For instance, Cisco s EtherChannel [21] product provides bandwidth aggregation across multiple Ethernet ....

D. C. Anderson, J. S. Chase, and A. M. Vahdat, "Interposed request routing for scalable network storage," in Proceedings of the Fourth Symposium on Operating System Design and Implementation (OSDI), October 2000.

....2. Data Striping Alternatives in the problems discussed in detail below, primarily because congestion control is not performed independently for each network path. Bandwidth aggregation has also been considered in the context of storage systems and high volume data or file servers, for instance [16], 17] 18] 19] Typical storageserver architectures are confined to be within a LAN having a single controlling authority. In this case bandwidth can be aggregated at the MAC layer. For instance, Cisco s EtherChannel [20] product provides bandwidth aggregation across multiple Ethernet links. ....

D. C. Anderson, J. S. Chase, and A. M. Vahdat, "Interposed request routing for scalable network storage," in Proceedings of the Fourth Symposium on Operating System Design and Implementation (OSDI), October 2000.

....shrinks these packets by removing all dummy data (except for the DLIs) from the payload of the packets, retaining all other information such as the IP headers, UDP TCP headers and the application reply headers. These shrunk packets are then redirected to the NADs. Note that unlike in Slice [4] where packets are redirected by manipulating packet headers, shrunk packets in SPIRAL are actually tunneled to the NADs with the packet headers treated as normal payload. Tunneling the packets simplifies the receiving code on the NADs and allows us to leverage the existing security protection ....

....However, LARD assumes a cluster configuration where each back end server is capable of serving requests independently. For NADs, such a scheme requires porting file system and applications (such as NFS HTTP server) on the disks, which essentially turns each NAD into a small server. In Slice [4], a request switching filter interposed along the network path between the client and the storage server routes file requests based on request type and parameters. Slice focuses on NFS over UDP and requires changes to the network routing components between the client and the server. Both the ....

D. Anderson, J. Chase, and A. Vahdat. Interposed request routing for scalable network storage. In Proc. of the 4th Symp. on OSDI, Oct. 2000.

....2.5 Related work There has been a huge amount of work in distributed file systems. Here, we focus on particularly relevant categories of related work. Layered clustering builds on the proxy concept [19] using interpositioning to add clustering to an existing client server protocol. Slice [1] uses this same concept to transparently provide unmodified clients access to scalable network storage. Layered clustering seeks to change neither the clients nor the servers, aiming for a subset of the design space with minimal changes to existing infrastructure. The fundamental consequence of ....

....stored) and the surrogate server (possibly multiple servers that store replicas of a data item) The interposition agent, that redirects requests and responses, can reside anywhere on the network path between the clients and servers. As such, it could be a part of the client protocol stack [1], a component of an intervening network element (e.g. the clustering switch of Figure 1) or a part of the server protocol stack. The forwarding of requests to surrogate servers could be done by re writing the original packet and placing it on the network, or by encapsulating the original ....

Darrell C. Anderson, Jeffrey S. Chase, and Amin M. Vahdat. Interposed request routing for scalable network storage. Symposium on Operating Systems Design and Implementation (San Diego, CA, 22--25 October 2000.

....LAN performance have created an opportunity to take a similar approach using a general purpose LAN as the storage backbone. Such systems are built from disks that are distributed throughout the network and attached to dedicated servers, cooperating peers, or the network itself. As pointed out in [3], they can be classified into two broad groups. One group of scalable storage systems, e.g. Frangipani Petal [14] layers the file system functions above a self manageable network storage volume using a shared disk model. Policies for data striping, redundancy, and storage site selection are done ....

....the value of and requirements for Object Oriented Devices in support of Network Attached Storage. Further, the American National Standard [1] has developed a command set proposal for Object based Storage Devices. In related work on development of self managed storage systems, the SLICE project [3] allows clients to virtualize storage, besides allowing direct access to the disks for large I O. This is achieved by introducing a request switching filter at the client s network path to the storage subsystem. The AutoRAID project [15] provides two levels of storage. The upper level has two ....

[Article contains additional citation context not shown here]

Anderson, D., Chase, J., and Vahdat, A. Interposed Request Routing for Scalable Network Storage. In Proceedings of OSDI, 2000.

.... live protocol [6] to maintain coherency between the proxy and server, which is also quite di#erent from the kind of consistency quality Data Staging clients are expecting. There are a number of studies that have been done with NFS as the underlying file system, most notably SFS [5] and Slice [3]. SFS is self certifying file system that provides secure, scalable access to public, read only data. The SFS client intercepts the NFS client requests and forwards it to the SFS server for security check. Similarly, Slice intercepts NFS request tra#c and distributes it across a server ensemble to ....

Anderson, D. C., Chase, J. S., and Vahdat, A. M. Interposed Request Routing for Scalable Network Storage. ACM Transactions on Computer Systems (February 2002)

....of these systems assume or require any knowledge of file system structures. When storage system interfaces are more developed than that provided in the local setting, there are more opportunities for new functionality. The use of a network packet filter within the Slice virtual file service [3] allows Slice to interpose on NFS traffic in clients, and thus implement a range of optimizations (e.g. preferential treatment of small files) Interposing on an NFS traffic stream is simpler than doing so on a SCSI disk block stream because the contents of NFS packets are well defined. High end ....

D. Anderson, J. Chase, and A. Vahdat. Interposed Request Routing for Scalable Network Storage. Transactions on Computer Systems (TOCS), 20(1), February 2002.

....component in data intensive applications such as multimedia databases, web information services, and data archival repositories. For these applications, their demand for storage capacity keeps growing as faster computers with larger memory at lower cost become available for more advanced computing [2, 8, 20, 22, 36]. To provide easy manageability, scattered storage resources need be unified and interfaced as a virtual disk, and component failures should be invisible to applications as much as possible. A popular solution is to interconnect multiple storage devices through a dedicated storage area network ....

....disk group be reorganized manually. Techniques such as hardware based dedicated hot spare or software based distributed spare lessens the burden of administration, but the process is still not transparent to administrators. Cluster based storage systems have gained attention from both academia [2, 4, 7, 12, 14, 15, 16] and industry [21, 34] as an alternative to provide storage for data intensive applications due to their scalability and cost advantage. In such a system, storage resources are directly attached to dedicated or general purpose cluster nodes. Cluster nodes coordinate with each other through a ....

[Article contains additional citation context not shown here]

ANDERSON, D., CHASE, J., AND VAHDAT, A. Interposed request routing for scalable network storage. In Proceedings of the Fourth Symposium on Operating System Design and Implementation (OSDI) (October 2000).

....finer granularity. FARSITE [1] replicates at the unit of a directory group , which resembles a volume, but with a dynamically defined boundary. It supports file renaming across directory groups using a Byzantine fault tolerant consensus protocol that coordinates nodes in a lock step manner. Slice [2] replicates files and directories independently over a cluster of servers and uses two phase commits to coordinate nodes. Pangaea, in contrast, coordinates nodes optimistically to improve availability and performance in a wide area, but it must detect and resolve conflicting updates. Data ....

....policy chooses gold replica sets for each file or directory uniformly randomly. The dir policy chooses gold replica sets uniformly randomly for a directory, but for regular files in the directory, it chooses the set the same as the directory s. This policy, similar to Archipelago s [5] and Slice s [2], helps directories to be concentrated on fewer nodes and lower the space overhead. Average number of bronze replicas per file. Bronze replicas impose the same storage overhead as gold replicas. Bronze replicas, however, are created only when the users wants to access it, and we can expect ....

Darrell C. Anderson, Jeffrey S. Chase, and Amin M. Vahdat. Interposed Request Routing for Scalable Network Storage. In 4th Symp. on Op. Sys. Design and Impl. (OSDI), pages 259--272, San Diego, CA, USA, October 2000.

....services. Many of the lessons learned in those systems apply to SRPC; for example, some of the techniques used in VINO to survive misbehaving kernel extensions are directly applicable to our framework. Slice is a virtual file service that extends services on only the client side via interposition [3]. By introducing client side packet filters, Slice can build a virtual file service on top of existing protocols such as NFS, and does so transparently to file system clients. Thus, Slice s interposition and SRPC server side scripting are complimentary approaches, with Slice adding activity to ....

D. Anderson, J. Chase, and A. Vahdat. Interposed Request Routing for Scalable Network Storage. Transactions on Computer Systems (TOCS), 20(1), February 2002.

....assume any knowledge of which blocks contain what type of file system data. When storage system interfaces are more developed than that provided in the local setting, there are more opportunities for new functionality. The use of a network packet filter within the Slice virtual file service [3] allows Slice to interpose on NFS traffic in clients, and thus implement a range of optimizations (e.g. preferential treatment of small files) Interposing on an NFS traffic stream is simpler than doing so on a local disk block stream because the contents and fields of NFS packets are ....

D. Anderson, J. Chase, and A. Vahdat. Interposed Request Routing for Scalable Network Storage. Transactions on Computer Systems (TOCS), 20(1), February 2002.

....follow the developments in cluster networking technologies and to share resources with workstations, whereas the low cost mass production requirements for network attached disks may restrict the offered functionality. Using regular nodes as storage nodes has previously been suggested in the Slice [3, 8] and OPIOM [13] projects, but where they primarily focus on using dedicated storage nodes, we examine the possibilities for distributing the load across all nodes in a cluster. Petal [23] is a fault tolerant, extensible, virtual disk based on dedicated storage nodes. Proboscis could complement ....

D. C. Anderson, J. Chase, and A. Vahdat. Interposed request routing for scalable network storage. In Proc. of the Fourth Symposium on Operating Systems Design and Implementation, Oct. 2000.

....In this example, the local file system of node 2 is part of two federated file systems. A1 runs only on node 1 and uses the local file system directly. There is a significant body of research related to distributed file systems [4, 5, 3, 6, 7, 8, 9, 10, 1, 2, 11, 12] Some recent projects include [13], the emerging industry standard DAFS [14] and wide area systems like [16, 17, 18, 19] In our project, we combine two technologies: the federated file system idea, and the remote memory communication support. Remote memory communication is the key ingredient of the recently proposed system area ....

Darrell C. Anderson, Je#rey S. Chase, Amin M. Vahdat. Interposed Request Routing for Scalable Network Storage . Proc. of 4th OSDI (2000).

....system. NFS 2 is a protocol level service and can leverage diverse file systems for optimal content placement and delivery. Nevertheless, NFS 2 is complementary to cluster file systems a partition can be implemented as a cluster file system and can be integrated into a broader file space. Slice [6] is a system that also uses a partitioning approach, similar to NFS 2. Slice s file placement policies (small versus large files and a deterministic distribution within each class of files) are implemented in tproxies modules that forward client operations to the right partition, operating at the ....

Anderson, D., Chase, J., and Vadhat, A., Interposed Request Routing for Scalable Network Storage, in Proc. of the USENIX OSDI, San Diego, CA, USA, October, 2000.

....connection across multiple IP links. Thus the multipath QoS routing results found in the literature can be used in conjunction with a connection striping scheme. Bandwidth aggregation has also been considered in the context of storage systems and high volume data or file servers, for instance [17], 18] 19] 20] Typical storage server architectures are confined to within a LAN having a single controlling authority. Note that Within a LAN, it is possible to do bandwidth aggregation at the MAC layer. For instance, Cisco s EtherChannel [21] product provides bandwidth aggregation across ....

Darrell C. Anderson, Jeffrey S. Chase, and Amin M. Vahdat, "Interposed request routing for scalable network storage," in Proceedings of the Fourth Symposium on Operating System Design and Implementation (OSDI), October 2000.

....then shrinks these packets by removing all dummy data (except for the DLIs) from the payload of the packets, retaining all other information such as the IP headers, UDP TCP headers, and the application reply headers. These shrinked packets are then redirected to the disks. Note that unlike in [2] where packets are redirected by manipulating packet headers, shrinked packets in SPIRAL are actually tunneled to the disks with the packet headers treated as normal payload. Tunneling the packets simplifies the receiving code on the disks and allows us to leverage the existing security protection ....

....turning each disk into a small server. An extension of LARD which supports persistent HTTP connection is presented in [3] A back end forwarding mechanism which forwards replies among the back end nodes is proposed to avoid the complexity and overhead of multiple TCP handoffs. In Slice [2], a request switching filter interposed along the network path between the client and the storage server routes file requests based on request type and parameters. Slice focuses on NFS over UDP and requires changes to the network routing components between the client and the server. Both the ....

D. Anderson, J. Chase, and A. Vahdat. Interposed request routing for scalable network storage. Proc. of 4th Symp. OSDI, 2000.

....the developments in cluster networking technologies and to share resources with workstations, whereas the low cost mass production requirements for network attached disks will possibly restrict the o ered functionality. However, the ideas on adapting le system structure to network attached disks [21], 22] can be applied to a Proboscis based storage area network as well. Using regular nodes as storage nodes has previously been suggested in the Slice [21] 23] and OPIOM [24] projects, but where they primarily focus on using dedicated storage nodes, we examine the possibilities for ....

....disks will possibly restrict the o ered functionality. However, the ideas on adapting le system structure to network attached disks [21] 22] can be applied to a Proboscis based storage area network as well. Using regular nodes as storage nodes has previously been suggested in the Slice [21], 23] and OPIOM [24] projects, but where they primarily focus on using dedicated storage nodes, we examine the possibilities for distributing the load across all nodes in a cluster. Our I O bu er management scheme complements existing work on global memory management [25] 26] and cooperative ....

D. C. Anderson, J. Chase, and A. Vahdat, \Interposed request routing for scalable network storage," in Proceedings of the Fourth Symposium on Operating Systems Design and Implementation, Oct. 2000.

....nodes to handle all client sessions. This approach can be seen as a software only version of the SAN based architecture. Using this architecture in data intensive services requires the file system to be highly scalable under changing workload and system configurations. Such file systems do exist [6, 7, 150], but they are still in the basic research stage due to their sheer complexity. Even if they were available now, their level of manageability and availability would not be optimal, because they offer generic, single copy semantics and sacrifice availability. For example, they tolerate only a ....

Darrell C. Anderson, Jeffrey S. Chase, and Amin M. Vahdat. Interposed Request Routing for Scalable Network Storage. In 4th Symp. on Operating System Design and Implementation (OSDI), pages 259--272, San Diego, CA, October 2000. 1.3.5

....a single connection across multiple IP links. Thus the multi path QoS routing results in the literature can be used in conjunction with a connection striping scheme. Bandwidth aggregation has also been considered in the context of storage systems and high volume data file servers (for instance [15], 16] 17] 18] etc. Typical storage server architecture is confined to be within a LAN with a single controlling authority. Note that Within a LAN, it is possible to do bandwidth aggregation at the MAC layer. For instance, Cisco s EtherChannel [19] product provides bandwidth aggregation ....

Darrell C. Anderson, Jeffrey S. Chase, and Amin M. Vahdat, "Interposed request routing for scalable network storage," in Proceedings of the Fourth Symposium on Operating System Design and Implementation (OSDI), October 2000.

....not necessarily the same with their children in the namespace, as depicted in Figure 1; a file with inode number 1001, which resides in partition 1, is referenced with the name passwd from a directory in partition N. Other systems that follow a similar approach include Slice from Duke University [1] and Archipelago from Princeton [9] While the intention of this report is to investigate protocols for building robust namespaces in the context of DiFFS, the problem is more generic. It can be broadly stated as: maintaining a consistent namespace over a collection of distributed objects ....

....of the above operations potentially spans more than one site in a distributed system. The site containing the directory (namespace object) and the one containing the referenced object can be physically apart. Slice and Archipelago use 2 phase commit to implement distributed namespace operations [1, 9]. Atomic commitment protocols are known to have a high computational cost [16, 5] They impose a high overhead to failure free execution, due to synchronous logging in the critical path of the operations. Additionally, they lock system resources across all the sites involved in the protocol for ....

[Article contains additional citation context not shown here]

Anderson, D., Chase, J., and Vadhat, A. "Interposed Request Routing for Scalable Network Storage", in Proc. of the Usenix OSDI. San Diego, CA, USA.

....systems, such as Coda [19] and Echo [8] These systems are very concerned with maintaining wide area consistency, but make optimistic assumptions when the network is not available. Latency hiding has been explored with the concept of buffer servers [5] particularly in the use of microproxies [4] to intercept and buffer NFS operations. It has been suggested that Kangaroo bears a certain similarity to peer to peer file sharing systems, such as Gnutella or Freenet [10] Although the interface is similar a client may perform I O from any node in a cloud of Kangaroo servers the naming ....

D. Anderson, J. Chase, and A. Vahdat. Interposed request routing for scalable network storage. In Proceedings of the Fourth Symposium on Operating Systems Design and Implementation, 2000.

....virtualize these services. In particular, storage is increasingly network based, and network storage architectures are shifting from FibreChannel to IP networks. Server switching technology could enable scalable virtual servers for IP based network storage protocols such as iSCSI and NFS (e.g. [2]) This paper discusses the interplay of these factors, and speculates on their implications. These factors suggest a more limited role for sophisticated contentbased switching features in the future, and an expanding role for server switches as a focal point for managing server resources rather ....

....competitive with FibreChannel from a price performance standpoint, and will soon advance with a new crop of devices based on the emerging iSCSI block storage standard. All of these factors are feeding demand for scalable IP based network storage solutions. In our work on the Slice storage system [2, 1] we are using server switching techniques to construct a scalable virtual storage appliance . Slice uses content routing to distribute file service traffic across a dynamic ensemble of servers and network attached block storage devices. An important goal of Slice is to automatically balance the ....

Darrell C. Anderson, Jeffrey S. Chase, and Amin M. Vahdat. Interposed request routing for scalable network storage. In Proceedings of the Fourth Symposium on Operating System Design and Implementation (OSDI), October 2000.

....service [3, 5, 7] These forwarding agents inspect more deeply into the packet than do routers, which are typically This work is supported by the National Science Foundation (EIA9870724 and EIA 9972879) Intel Corporation, and Myricom. concerned with the IP header. Host based cluster frontends [8, 1] enable application specific forwarding functionality [9] but can be limited by the PCI bus in common commodity host architectures. The two trips across the I O bus (upon reception and transmission) halve potential forwarding bandwidth when the bus and network speeds are equal. Payload caching ....

Darrell Anderson, Jeff Chase, and Amin Vahdat. Interposed request routing for scalable network storage. In Fourth Symposium on Operating Systems Design and Implementation, 2000.

....[9] acts as a a fundamental enabler for flexible resource provisioning, by automatically directing request traffic toward selected servers. We are pleased that these projects have yielded recent high profile publications in OSDI and SOSP respectively the major conferences in my field. Slice [3, 2] is a decentralized storage service architecture intended to meet the I O needs of Internet server clusters as well as data intensive computing. Demand for large scale storage services is growing rapidly along with the Web, driven not just by Web services but also by new content capture devices ....

D. C. Anderson, J. S. Chase, and A. M. Vahdat. Interposed request routing for scalable network storage. In Proceedings of the Fourth Symposium on Operating System Design and Implementation (OSDI), October 2000. Award paper selected for a special issue of ACM Transactions on Computer Systems (TOCS).

....Although more data is needed, the results justify an expected potential overall energy savings of 25 or more for Web data centers. 7. RELATED WORK Cluster based network services. Many systems have used server clusters and redirecting front ends or switches to virtualize network services (e.g. [32, 36, 7, 4]) for scalability, availability, or both. Muse uses reconfigurable redirecting switches as a mechanism to support adaptive resource provisioning in network server pools; related projects considering this approach are DDSD [47] and Oceano [5] The switch features needed are similar to the support ....

Darrell C. Anderson, Jeffrey S. Chase, and Amin M. Vahdat. Interposed Request Routing for Scalable Network Storage. In Proceedings of the Fourth Symposium on Operating System Design and Implementation (OSDI), October 2000.

....a Web server and one or more proxy caches. Other examples of forwarding intermediaries include firewalls, content routers, protocol converters [7] network address translators (NAT) and overcast multicast nodes [10] New forwarding intermediaries are introduced in the network storage domain [12, 1], Web services [9] and other networked data delivery. This paper investigates a technique called payload caching to improve data forwarding performance on intermediaries. In this paper, we define forwarding as the simple updating of packet headers and optional inspection of data as it flows ....

Darrell Anderson, Jeff Chase, and Amin Vahdat. Interposed request routing for scalable network storage. In Fourth Symposium on Operating Systems Design and Implementation, 2000.

No context found.

ANDERSON,D.,CHASE,J.,AND VAHDAT,A. 2000. Interposed request routing for scalable network storage. In Proceedings of the Symposium on Operating Systems Design and Implementation.

No context found.

D. Anderson, J. Chase, and A. Vahdat. Interposed request routing for scalable network storage. In Proceedings of the Symposium on Operating Systems Design and Implementation, 2000.

No context found.

D. Anderson, J. Chase, and A. Vahdat. Interposed request routing for scalable network storage. In Proceedings of the Symposium on Operating Systems Design and Implementation, 2000.

No context found.

D. Anderson, J. Chase, and A. Vahdat. Interposed request routing for scalable network storage. In Proceedings of the Fourth Symposium on Operating Systems Design and Implementation, 2000.

No context found.

D. C. Anderson, J. S. Chase, and A. M. Vahdat. Interposed request routing for scalable network storage. In Proceedings of the 4th Symposium on Operating Systems Design and Implementation (OSDI), Oct. 2000.

No context found.

Anderson, D., Chase, J., and Vahdat, A. Interposed Request Routing for Scalable Network Storage. In Proceedings of OSDI, 2000.

No context found.

D. Anderson, J. Chase, and A. Vahdat. Interposed request routing for scalable network storage. In Proceedings of the Fourth Symposium on Operating Systems Design and Implementation, 2000.

No context found.

D. Anderson, J. Chase, and A. Vahdat, Interposed request routing for scalable network storage, in Proceedings of the Fourth Symposium on Operating Systems Design and Implementation, 2000.

No context found.

D. C. Anderson, J. S. Chase, and A. M. Vahdat. Interposed request routing for scalable network storage. In Proceedings of the 4th Symposium on Operating Systems Design and Implementation (OSDI), Oct. 2000.

No context found.

D. Anderson, J. Chase, and A. Vahdat. Interposed Request Routing for Scalable Network Storage. ACM Transactions on Computer Systems, 20(1), February 2002.

No context found.

D. C. Anderson, J. S. Chase, and A. M. Vahdat. Interposed request routing for scalable network storage. In Proceedings of the 4th Symposium on Operating Systems Design and Implementation (OSDI), Oct. 2000.

No context found.

D. C. Anderson, J. S. Chase, and A. M. Vahdat. Interposed request routing for scalable network storage. In Proceedings of the 4th Symposium on Operating Systems Design and Implementation (OSDI), Oct. 2000.

No context found.

D. Anderson, J. Chase, and A. Vahdat. Interposed Request Routing for Scalable Network Storage. ACM Transactions on Computer Systems, 20(1), February 2002.

No context found.

. D. Anderson, J. Chase, A. Vahdat, "Interposed Request Routing for Scalable Network Storage", in Proc. of 4th USENIX OSDI Symposium, pp. 259-272, San Diego, CA, October 2000.

No context found.

D. Anderson, J. Chase, and A. Vahdat. Interposed request routing for scalable network storage. In Proceedings of the Fourth Symposium on Operating Systems Design and Implementation, 2000.

No context found.

D. C. Anderson, J. S. Chase, and A. M. Vahdat. Interposed request routing for scalable network storage. In Proc. of the Fourth USENIX Symposium on Operating Systems Design and Implementation (OSDI), pages 259--272, Oct. 2000.

No context found.

Darrell C. Anderson, Jeffrey S. Chase, and Amin M. Vahdat. Interposed request routing for scalable network storage. ACM Transactions on Computer Systems, 20(1):25--48. ACM Press, February 2002.

No context found.

ANDERSON, D., CHASE, J., AND VAHDAT, A. Interposed Request Routing for Scalable Network Storage. In Proc. of Operating Systems Design and Implementation (2000).

No context found.

Darrel C. Anderson, Jerey S. Chase, and Amin M. Vahdat, Interposed request routing for scalable network storage, In Proc. of the Fourth USENIX Symposium on Operating Systems Design and Implementation (OSDI), pp. 259272, Oct. 2000.

No context found.

D. Anderson, J. Chase, and A. Vahdat, Interposed request routing for scalable network storage, in Proceedings of the Fourth Symposium on Operating Systems Design and Implementation, 2000.

No context found.

D. Anderson, J. Chase, and A. Vahdat. Interposed request routing for scalable network storage. In Proceedings of the Fourth Symposium on Operating Systems Design and Implementation, 2000.

No context found.

D. C. Anderson, J. S. Chase, and A. M. Vahdat. Interposed request routing for scalable network storage. ACM Transactions on Computer Systems, 20(1):25--48. ACM Press, February 2002.

No context found.

Darrell C. Anderson, Jeffrey S. Chase, and Amin M. Vahdat. "interposed Request Routing for Scalable Network Storage," in Proceedings of the Fourth Symposium on Operating System Design and Implementation (OSDI), October 2000.

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