公开(公告)号：US06625750B1

公开(公告)日：2003-09-23

申请号：US09440756

申请日：1999-11-16

申请人： Wayne W. Duso ,  Leslie E. Kuczynski ,  John Forecast ,  Uday Gupta ,  Uresh K Vahalia ,  Dennis P. J. Ting

发明人： Wayne W. Duso ,  Leslie E. Kuczynski ,  John Forecast ,  Uday Gupta ,  Uresh K Vahalia ,  Dennis P. J. Ting

IPC分类号： H02H305

CPC分类号： G06F11/2038 ,  G06F11/2028 ,  H04L41/0681 ,  H04L43/0817

摘要： A file server includes a plurality of stream server computers linking data storage to a data network, and at least two controller servers for controlling the stream server computers. The controller servers are programmed so that at a given time one of the controller servers is active in controlling the stream server computers, and another of the controller servers is inactive. The inactive controller server is programmed to respond automatically to a failure of the active controller server by becoming active. For example, each of the controller servers has a respective flag for indicating whether or not the controller server is active. Each controller server is programmed so that, upon booting, it will read the flag of the other stream server, and if the flag of the other controller server indicates that the other controller server is active, then the controller server becomes inactive. Otherwise, the stream server assumes an active or inactive state based on a predetermined arbitration method. The active controller server also reports failure conditions to a customer service center, monitors the inactive controller server to ensure that the inactive controller server is in a state of readiness, and monitors itself to determine whether it should become inactive.

摘要翻译： 文件服务器包括将数据存储器连接到数据网络的多个流服务器计算机以及用于控制流服务器计算机的至少两个控制器服务器。 控制器服务器被编程，使得在给定时间，一个控制器服务器在控制流服务器计算机中是活动的，另一个控制器服务器处于非活动状态。 无效控制器服务器被编程为通过变为活动来自动响应主动控制器服务器的故障。 例如，每个控制器服务器具有用于指示控制器服务器是否活动的相应标志。 每个控制器服务器被编程，使得在引导时，它将读取另一个流服务器的标志，如果其他控制器服务器的标志指示其他控制器服务器是活动的，则控制器服务器变为不活动。 否则，流服务器基于预定的仲裁方法假定为活动或非活动状态。 主动控制器服务器还向客户服务中心报告故障状况，监视非活动控制器服务器以确保非活动控制器服务器处于准备状态，并监控自身以确定其是否应变为非活动状态。

公开(公告)号：US5987621A

公开(公告)日：1999-11-16

申请号：US851507

申请日：1997-05-05

申请人： Wayne W. Duso ,  John Forecast ,  Uday Gupta ,  Uresh K Vahalia ,  Dennis P. J. Ting

发明人： Wayne W. Duso ,  John Forecast ,  Uday Gupta ,  Uresh K Vahalia ,  Dennis P. J. Ting

IPC分类号： G06F11/14 ,  G06F11/20 ,  H04N21/232 ,  H04N21/2387 ,  H04N21/239 ,  H04N21/24 ,  G06F11/00

CPC分类号： G06F11/2046 ,  G06F11/2023 ,  G06F11/2035 ,  H04N21/232 ,  H04N21/2387 ,  H04N21/2396 ,  H04N21/2404 ,  G06F11/1471

摘要： A file server includes stream server computers linking a cached disk array storage subsystem to a data network, and at least two controller servers for receiving requests for file access from network clients. At any given time one of the controller servers is active and another is inactive in servicing client requests. The active controller server selects one of the stream servers to service each request. A controller server failover mechanism is provided for recovering from a failure of the active controller server, and a stream server failover mechanism is provided for recovering from a failure of a stream server. The inactive controller server becomes active when it fails to receive a signal periodically transmitted by the active controller server. The active controller server begins stream server failover when it fails to receive a signal periodically transmitted by each stream server. To resume automatically an interrupted task, the tasks are organized as a series of transactions, and each transaction includes operations which can be duplicated without substantial disruption. The active controller server commits results of each transaction to memory of the cached disk array. Before becoming active, the inactive controller recovers the committed state of the interrupted tasks from the cached disk array.

摘要翻译： 文件服务器包括将缓存的磁盘阵列存储子系统链接到数据网络的流服务器计算机，以及用于从网络客户端接收文件访问请求的至少两个控制器服务器。 在任何给定的时间，其中一个控制器服务器是活动的，另一个在服务客户机请求时处于不活动状态 主动控制器服务器选择一个流服务器来为每个请求提供服务。 提供控制器服务器故障切换机制用于从主动控制器服务器的故障中恢复，并且提供流服务器故障切换机制用于从流服务器的故障中恢复。 无效控制器服务器在接收到主动控制器服务器周期性发送的信号失败时变为活动状态。 当主控制器服务器无法接收每个流服务器周期性发送的信号时，开始流服务器故障切换。 要自动恢复中断的任务，这些任务被组织成一系列事务，每个事务包括可重复的操作，而不会发生大的中断。 主动控制器服务器将每个事务的结果提交给缓存磁盘阵列的内存。 在活动之前，非活动控制器从缓存的磁盘阵列中恢复已中断任务的提交状态。

公开(公告)号：US08417788B1

公开(公告)日：2013-04-09

申请号：US09966709

申请日：2001-09-28

申请人： Dennis P. J. Ting ,  Uresh K. Vahalia ,  John A. Hilliar ,  Uday Gupta

发明人： Dennis P. J. Ting ,  Uresh K. Vahalia ,  John A. Hilliar ,  Uday Gupta

IPC分类号： G06F15/167

CPC分类号： G06F17/30233 ,  H04L12/4641

摘要： A file system that utilizes Virtual Local Area Network (VLAN) addressing and routing schemes to permit implementing multiple networked file system applications in a single data processing system. In particular, a networked file server or data mover has associated with it a mass storage device such as one or more disk drives, as well as one or more Network Interface Cards (NICs). A network interface to the data mover is associated with VLAN identifiers (IDs) such that a VLAN ID can be assigned to each file system. At the physical layer of the data mover, VLAN information is encapsulated and passed up to various higher protocol layers, such as a network layer and/or a transport layer, eventually to a selected file system as dictated by the VLAN ID. The file system application in effect becomes an addressable element of the VLAN. A single data mover can thus be deployed to service multiple groups of end users, as long as the members of each group have different VLAN identifiers for the multiple file systems. The result provides security, access restrictions, and all the other benefits of VLAN protocols that would be obtained by deploying separate data movers for each VLAN, without incurring the acquisition expense and administrative overhead associated with having to purchase and maintain multiple data mover systems.

摘要翻译： 一种使用虚拟局域网（VLAN）寻址和路由方案来允许在单个数据处理系统中实现多个联网文件系统应用程序的文件系统。 特别地，联网的文件服务器或数据移动器与其相关联的大容量存储设备，例如一个或多个磁盘驱动器以及一个或多个网络接口卡（NIC）。 与数据移动设备的网络接口与VLAN标识符（ID）相关联，使得可以为每个文件系统分配VLAN ID。 在数据移动设备的物理层，VLAN信息被封装并传递到各种较高的协议层，如网络层和/或传输层，最终到VLAN ID所指定的所选文件系统。 文件系统应用程序有效成为VLAN的可寻址元素。 因此，只要每个组的成员对多个文件系统具有不同的VLAN标识符，就可以部署单个数据移动器来服务多个终端用户组。 该结果提供了通过为每个VLAN部署单独的数据移动器而获得的VLAN协议的安全性，访问限制和所有其他优点，而不会导致与必须购买和维护多个数据移动器系统相关的采购费用和管理开销。

公开(公告)号：US06298386B1

公开(公告)日：2001-10-02

申请号：US08747769

申请日：1996-11-13

申请人： Uresh K Vahalia ,  Uday Gupta ,  Aju John ,  Dennis P. J. Ting ,  Percy Tzelnic

发明人： Uresh K Vahalia ,  Uday Gupta ,  Aju John ,  Dennis P. J. Ting ,  Percy Tzelnic

IPC分类号： G06F1516

CPC分类号： G06F17/30203

摘要： There is a performance loss associated with servicing a pipe or stream for a connection oriented process by maintaining a connection between a server thread and a client for a series of messages. As a result of maintaining this connection, there is less balance; some threads work harder than others, causing a loss of performance. To solve this problem, a collector queue combines messages from the connection oriented process with messages from the other concurrent processes. The threads receive messages from the collector queue rather than individual pipes. Any idle thread can pick up a message from the collector queue. The collector queue keeps track of which pipe each message came from so that the reply of the server to each message is directed to the same pipe from which the message came from. Therefore the collector queue ensures thread balance and efficiency in servicing the messages. In the preferred implementation, each entry in the collector queue includes a message pointer and a pipe pointer. The message pointer points to allocated memory storing the message in a message buffer. The pipe pointer points to the pipe from which the message originated. The collector queue is a singly linked list. A free thread takes an entry off the collector queue, interprets the message of the entry, sends a reply, and deallocates the memory of the entry and the allocated memory storing the message in the message buffer.

摘要翻译： 通过维护一系列消息的服务器线程和客户端之间的连接，为为面向连接的进程维护管道或流而导致性能损失。 由于维持这种联系，平衡较少; 一些线程比其他线程更难工作，导致性能下降。 为了解决这个问题，收集器队列将来自面向连接进程的消息与来自其他并发进程的消息相结合。 线程从收集器队列接收消息，而不是单个管道。 任何空闲线程都可以从收集器队列中接收到一条消息。 收集器队列跟踪每个消息来自哪个管道，使得服务器对每个消息的回复被定向到来自消息的相同管道。 因此，收集器队列可确保线程平衡和服务消息的效率。 在优选实现中，收集器队列中的每个条目包括消息指针和管道指针。 消息指针指向在消息缓冲器中存储消息的分配的存储器。 管道指针指向消息始发的管道。 收集器队列是单链表。 一个空闲的线程从收集器队列中取消入口，解释该条目的消息，发送回复，并将该条目的存储器和存储消息的分配的存储器释放到消息缓冲器中。

公开(公告)号：US06275953B1

公开(公告)日：2001-08-14

申请号：US08938859

申请日：1997-09-26

申请人： Uresh K. Vahalia ,  Uday Gupta ,  Dennis P. J. Ting

发明人： Uresh K. Vahalia ,  Uday Gupta ,  Dennis P. J. Ting

IPC分类号： G06F1116

CPC分类号： G06F11/0793 ,  G06F11/0709 ,  G06F11/2007 ,  G06F11/203 ,  G06F11/2035 ,  G06F11/2038 ,  G06F11/2046 ,  H04L41/0654 ,  H04L43/0817

摘要： A network file server includes a first set of data processors for receiving requests from clients, and a second set of data processors for accessing read-write file systems. A respective data processor in the second set is assigned to each file system for exclusive management of locks on the file system. The file server can detect failure of a failed data processor and automatically recover from the failure. When a failure of a data processor in the first set is detected, a spare data processor is programmed with the logical and physical network addresses of the failed data processor so that the spare data processor assumes the network identity of the failed data processor. When a failure of a data processor in the second set is detected, responsibility for management of the locks on each file system managed by the failed data processor is transferred to an operational data processor. Preferably the responsibility is transferred to the operational data processors in such a way as to balance loading on the operational data processors. The data processors can be commodity digital computers for low cost, and a cached disk storage subsystem or file system caches and remote dual copy techniques can be used to ensure high performance and high data availability.

摘要翻译： 网络文件服务器包括用于接收来自客户端的请求的第一组数据处理器和用于访问读写文件系统的第二组数据处理器。 第二组中的相应数据处理器被分配给每个文件系统，以对文件系统上的锁进行专用管理。 文件服务器可以检测出故障的数据处理器的故障，并自动从故障恢复。 当检测到第一组中的数据处理器的故障时，利用故障数据处理器的逻辑和物理网络地址对备用数据处理器进行编程，使得备用数据处理器承担故障数据处理器的网络标识。 当检测到第二组中的数据处理器的故障时，由故障数据处理器管理的每个文件系统上的锁的管理责任被传送到操作数据处理器。 优选地，责任以将平衡在操作数据处理器上的负载的方式传送到操作数据处理器。 数据处理器可以是低成本的商品数字计算机，并且可以使用缓存的磁盘存储子系统或文件系统高速缓存和远程双重复制技术来确保高性能和高数据可用性。

公开(公告)号：US06499039B1

公开(公告)日：2002-12-24

申请号：US09404016

申请日：1999-09-23

申请人： Dinesh Venkatesh ,  Wayne W. Duso ,  John Forecast ,  Uday Gupta

发明人： Dinesh Venkatesh ,  Wayne W. Duso ,  John Forecast ,  Uday Gupta

IPC分类号： G06F1200

CPC分类号： G06F17/30067 ,  G06F3/0607 ,  G06F3/0632 ,  G06F3/0643 ,  G06F3/0683 ,  Y10S707/99952 ,  Y10S707/99953 ,  Y10S707/99955

摘要： To reorganize a striped file system, data blocks are sequentially moved from storage locations in original data storage to storage locations in an array including the original data storage and additional data storage. If the new striping has the same precedence order as the original striping, then temporary storage is not needed for the block movement. Otherwise, stripes of the data blocks are sequentially moved to a stripe of temporary storage locations, and moved from the stripe of the temporary locations to storage locations in the array. Once a pivot point is reached, however, there is no need to use the temporary storage. Moreover, there is an increasing separation between the read and write pointers. Once this separation exceeds the stripe length, the file system metadata can be synchronized at decreasing intervals and still permit concurrent read access. The number of block transfers between the synchronization points can increase exponentially, and the synchronization points can be aligned with extent boundaries.

摘要翻译： 为了重组条带文件系统，数据块从原始数据存储中的存储位置顺序地移动到包括原始数据存储和附加数据存储的阵列中的存储位置。 如果新条带与原始条纹具有相同的优先顺序，则块运动不需要临时存储。 否则，数据块的条带被顺序地移动到临时存储位置的条带，并且从临时位置的条带移动到阵列中的存储位置。 一旦达到枢纽点，就不需要使用临时存储。 此外，读指针和写指针之间存在越来越大的差距。 一旦这种分离超过条带长度，文件系统元数据可以以减小的间隔同步，并且仍然允许并发读取访问。 同步点之间的块传输次数可以指数增加，并且同步点可以与扩展边界对齐。

公开(公告)号：US06453354B1

公开(公告)日：2002-09-17

申请号：US09261626

申请日：1999-03-03

申请人： Xiaoye Jiang ,  Uresh K. Vahalia ,  Uday Gupta ,  Percy Tzelnic

发明人： Xiaoye Jiang ,  Uresh K. Vahalia ,  Uday Gupta ,  Percy Tzelnic

IPC分类号： C06F1516

CPC分类号： G06F17/30171

摘要： A first data mover computer services data access requests from a network client, and a second data mover computer is coupled to the first data mover computer for servicing data access requests from the first data mover computer. The first data mover computer uses a connection-oriented protocol to obtain client context information and to respond to a session setup request from the client by authenticating the client. Then the first data mover computer responds to a file system connection request from the client by forwarding the client context information and the file system connection request to the second data mover computer. Then the first data mover computer maintains a connection between the first data mover computer and the second data mover computer when the client accesses the file system and the first data mover computer passes file access requests from the client to the second data mover computer and returns responses to the file access requests from the second data mover computer to the client. In a preferred embodiment, the connection-oriented protocol is the Common Internet File System (CIFS) Protocol, and multiple clients share a Transmission Control Protocol (TCP) connection between the first data mover computer and the second data mover computer by allocation of virtual channels within the shared TCP connection and multiplexing of data packets of the virtual channels over the shared TCP connection.

摘要翻译： 第一数据移动器计算机服务来自网络客户端的数据访问请求，并且第二数据移动器计算机耦合到第一数据移动器计算机，用于为来自第一数据移动器计算机的数据访问请求提供服务。 第一数据移动器计算机使用面向连接的协议来获得客户端上下文信息并且通过认证客户端来响应来自客户端的会话建立请求。 然后，第一数据移动器计算机通过将客户端上下文信息和文件系统连接请求转发到第二数据移动器计算机来响应来自客户端的文件系统连接请求。 然后，当客户端访问文件系统并且第一数据移动器计算机将文件访问请求从客户端传递到第二数据移动器计算机时，第一数据移动器计算机维护第一数据移动计算机和第二数据移动计算机之间的连接，并且返回响应 将文件访问请求从第二数据移动计算机发送到客户机。 在优选实施例中，面向连接的协议是通用因特网文件系统（CIFS）协议，并且多个客户端通过分配虚拟信道来共享第一数据移动器计算机和第二数据移动器计算机之间的传输控制协议（TCP）连接 在共享TCP连接中并通过共享TCP连接复用虚拟通道的数据包。

公开(公告)号：US5948062A

公开(公告)日：1999-09-07

申请号：US747875

申请日：1996-11-13

申请人： Percy Tzelnic ,  Uresh K Vahalia ,  Dennis P.J. Ting ,  Uday Gupta

发明人： Percy Tzelnic ,  Uresh K Vahalia ,  Dennis P.J. Ting ,  Uday Gupta

IPC分类号： G06F11/14 ,  G06F12/08 ,  G06F17/30 ,  H04L29/06 ,  H04L29/08 ,  H04N7/173 ,  H04N7/14

CPC分类号： H04L67/42 ,  G06F11/1435 ,  G06F12/0804 ,  G06F12/0813 ,  G06F12/0815 ,  G06F12/0866 ,  G06F17/30171 ,  G06F17/30224 ,  H04L67/1095 ,  H04L67/2842 ,  H04N21/2182 ,  H04N21/23106 ,  H04N21/2396 ,  H04N21/47202 ,  H04N21/816 ,  H04N7/17318 ,  H04N7/17336 ,  G06F2212/312 ,  H04L29/06027 ,  H04L67/2895 ,  Y10S707/99953

摘要： For network file access, a plurality of data mover computers are linked to a cached disk array. Read-write files stored in the cached disk array are accessible to network clients through more than one of the data mover computers so that the data mover computers are not a bottleneck to file access. The cached disk array stores a network file directory including file locking information, and the data mover computers each have file system software that accesses the network file directory for shared read-write file access. The data mover computers perform the file system tasks such as managing the file directory for mapping of file names to logical blocks and for locking and unlocking of the files in order to minimize loading on the cached disk array. Therefore, the network file server need not use any particular one of the data mover computers to serve as a control point for read-write file access. Instead, file system software for managing the file directory is replicated in each data mover computer. The cached disk array recognizes logical block addresses, and the cached disk array has a limited internal locking facility ensuring that reading or writing to a logical block is an atomic operation. Preferably the data mover computers are commodity personal computers.

摘要翻译： 对于网络文件访问，多个数据移动器计算机链接到高速缓存的磁盘阵列。 存储在缓存磁盘阵列中的读写文件可通过多个数据移动计算机访问网络客户端，以便数据移动器计算机不是文件访问的瓶颈。 缓存的磁盘阵列存储包括文件锁定信息的网络文件目录，并且数据移动器计算机各具有访问网络文件目录以获得共享读写文件访问的文件系统软件。 数据移动器计算机执行文件系统任务，例如管理文件目录以将文件名映射到逻辑块，以及用于锁定和解锁文件，以最小化对高速缓存的磁盘阵列的加载。 因此，网络文件服务器不需要使用任何特定的数据移动计算机作为读写文件访问的控制点。 相反，用于管理文件目录的文件系统软件在每个数据移动计算机中被复制。 缓存的磁盘阵列识别逻辑块地址，缓存的磁盘阵列具有有限的内部锁定功能，确保读取或写入逻辑块是原子操作。 优选地，数据移动器计算机是商品个人计算机。

公开(公告)号：US5893140A

公开(公告)日：1999-04-06

申请号：US747631

申请日：1996-11-13

申请人： Uresh K Vahalia ,  Uday Gupta ,  Betti Porat ,  Percy Tzelnic

发明人： Uresh K Vahalia ,  Uday Gupta ,  Betti Porat ,  Percy Tzelnic

IPC分类号： G06F11/14 ,  G06F12/08 ,  G06F17/30 ,  H04L29/08 ,  G06F12/00

CPC分类号： G06F17/30171 ,  G06F11/1435 ,  G06F12/0804 ,  G06F12/0813 ,  G06F12/0866 ,  H04L67/1095 ,  Y10S706/908

摘要： A conventional network file server has a file system that permits file attributes and file data to be written in any order. The conventional network file server may also support an asynchronous write protocol, in which file attributes and file data need not be written to disk storage until a client sends a commit request. This asynchronous write protocol has a data security problem if the attributes are written before the data and the server crashes before completing the writing of the data to disk storage. This security problem is solved by adding a file system cache and following a protocol that writes the attributes to storage after writing the data to storage. For example, the attributes and data are stored in the file system cache and are not written down to storage until receipt of a commit request. When the commit request is received, the data is sent first from the file system cache to storage. Then the attributes are sent from the file system cache to storage. Then the file server acknowledges completion of the commit operation. In a preferred embodiment, storage is provided by an integrated cached disk array (ICDA) having a buffer cache and an array of disk drives, and the file system cache is distributed in a plurality of data mover computers interfaced to the ICDA. The addition of the file system cache to solve the security problem also reduces the burden on the buffer cache in the ICDA.

摘要翻译： 传统的网络文件服务器具有允许以任何顺序写入文件属性和文件数据的文件系统。 传统的网络文件服务器还可以支持异步写入协议，其中文件属性和文件数据不需要写入磁盘存储器，直到客户端发送提交请求。 如果在完成将数据写入磁盘存储器之前，数据之前写入的属性和服务器崩溃，则此异步写入协议有数据安全问题。 通过添加文件系统缓存并在将数据写入存储后遵循将属性写入存储的协议来解决此安全问题。 例如，属性和数据存储在文件系统缓存中，并且不会被写入存储，直到收到提交请求。 收到提交请求后，首先将数据从文件系统缓存发送到存储。 然后将属性从文件系统缓存发送到存储。 然后文件服务器确认完成提交操作。 在优选实施例中，通过具有缓冲高速缓存和磁盘驱动器阵列的集成缓存磁盘阵列（ICDA）提供存储，并且文件系统高速缓存分布在与ICDA接口的多个数据移动器计算机中。 加上文件系统缓存来解决安全问题，也减轻了ICDA缓冲区高速缓存的负担。

公开(公告)号：US07739379B1

公开(公告)日：2010-06-15

申请号：US10871600

申请日：2004-06-18

申请人： Uresh K. Vahalia ,  Uday Gupta ,  Betti Porat ,  Percy Tzelnic

发明人： Uresh K. Vahalia ,  Uday Gupta ,  Betti Porat ,  Percy Tzelnic

IPC分类号： G06F15/16 ,  G06F15/167 ,  G06F15/173

CPC分类号： G06F17/30171

摘要： A network file server includes a first set of data processors for receiving requests from clients, and a second set of data processors for accessing read-write file systems. A respective data processor in the second set is assigned to each file system for exclusive management of read and write locks on the file system. Each data processor in the first set can authorize access to file systems directly accessed by more than one data processor in the second set. Processing of a request for access that is authorized is continued by the data processor that is assigned to manage the locks on the file system to be accessed. The exclusivity of lock management eliminates cache coherency problems, and dynamic load balancing can be used to prevent the lock management from becoming a bottleneck to performance. A preferred embodiment includes a cached disk storage subsystem linked to data mover computers. Each data mover has authorization and ownership information for a plurality of file systems, and maintains a local cache of locking information for file systems that the data mover can directly access. The data mover receives a client request, and if access is authorized, checks whether it can directly access the file, and if not, forwards the request to another data mover.

摘要翻译： 网络文件服务器包括用于接收来自客户端的请求的第一组数据处理器和用于访问读写文件系统的第二组数据处理器。 第二组中的相应数据处理器被分配给每个文件系统，用于对文件系统上的读取和写入锁定进行排他性管理。 第一组中的每个数据处理器可以授权访问由第二组中的多个数据处理器直接访问的文件系统。 被授权的访问请求的处理由被分配用于管理要访问的文件系统上的锁的数据处理器继续。 锁管理的排他性消除了缓存一致性问题，动态负载平衡可以用来防止锁管理成为性能的瓶颈。 优选实施例包括链接到数据移动器计算机的高速缓存的磁盘存储子系统。 每个数据移动器具有用于多个文件系统的授权和所有权信息，并维护数据移动者可以直接访问的文件系统的锁定信息的本地缓存。 数据移动器接收客户端请求，如果访问被授权，则检查它是否可以直接访问该文件，如果不是，则将该请求转发给另一个数据移动器。