網路文件共享系統(CIFS)
2009/03/04-鐘聖雄
http://www.digitimes.com.tw/tw/dt/n/shwnws.asp?rnrnCnlID=10&Cat=&Cat1=&id=124179
網路文件共享系統(Common Internet File System；CIFS)原先被稱為伺服器訊息區塊(Server Message Block；SMB)，該技術使得網路上的機器得以共享文件、周邊機器(如印表機)、串接阜等資源，也提供通過授權的程序間溝通(Inter- Process Communication；IPC)功能。
有別於網路文件系統(Network File System；NFS)，CIFS的訊息格式並沒有固定長度，且多數訊息都包含無固定長度的數據，也使得協議的複雜度增加。一般來說，CIFS會包括1個32位元組的訊息頭，以及1個或多個長度不固定的訊息體。
CIFS 最初是由IBM的Barry Feigenbaum所開發，目的是為了將DOS上的本地端文件介面，改造成為可供分享的網路文件系統。後來，微軟(Microsoft)接手了該計畫， 為了修正第1版的DES錯誤，而使用了NTLM(NT Lan Manager)第2版，並將CIFS與微軟和3com共同合作的另1個網路管理程序結合，修改為當前的版本，並建立其分散式文件系統。也因為這樣的緣 故，CIFS多半是用在安裝微軟作業系統的機器上，但後來CIFS在經過Unix伺服器廠商重新開發後，也可以被用在連接Unix伺服器和Windows 作業系統的客戶端機器上，並執行檔案分享，以及共用PC周邊的功能。
CIFS最早是被設計成運行於NetBIOS協議上；換言之，CIFS是建立在NetBEUI、IPX/SPX以及TCP/IP協議成立的前提之上。後來，微軟在Windows 2000作業系統中，將CIFS改造為可以直接運行在TCP/IP之上。
CIFS 的作業方式，基本上就是大家熟悉的客戶端－伺服器(Client - Server)模式，也就是先由客戶端發出請求，再由伺服器回應；在CIFS協定中，有一部分就是用來協調客戶端對伺服器發出的請求，也就是說，客戶端是 藉由文件系統的中介，得以拜訪文件伺服器。在IPC部分，CIFS也允許利用網際網路功能，讓不同的局部子網路或虛擬私人網路(Virtual Private Network；VPN)得以互相連結，如此一來，不同的子網路之間就得以進行遠端協同作業，例如文件分享或是共用印表機等PC周邊設備功能，都是利用 CIFS協議運作。
CIFS最為人詬病的缺點，就是它會佔用過多頻寬，因為每1台子網路內的客戶端電腦，都必須要廣播其存在，才能共享服 務，但是CIFS本身並不使用廣播功能，這樣的問題實際上是NetBIOS的服務定位協議所造成的。在使用微軟Windows伺服器時，都是透過 NetBIOS來協議與定位，而NetBIOS則會定時向特定伺服器廣播某個服務的存在，這樣的作業方式在伺服器較少時，並不會佔用過多的頻寬，但在伺服 器超過一定數目時，網路大塞車的情況則會愈來愈明顯。
網路塞車會對CIFS的使用者帶來重大的影響，特別是在伺服器間的網路延遲時間過長時，透過網際網路拜訪VPN，進行文件分享與協同作業時，延遲會更加明顯。
為 了改善頻寬問題，一般來說可以利用Windows Internet Name Service(WINS)定位協議緩和這樣的情況，也就是使用更高階的系統將服務集中化，但是這也相對地會造成訊息與網路保管問題。此外，使用動態網域 名稱伺服器(Dynamic Domain Name Server；Dynamic DNS)也可以解決CIFS的問題。
CIFS伺服 器可以向網路上的客戶端電腦提供文件系統與各種資源，但客戶端卻可以保留分享權限，這使得CIFS得到廣泛的採用。此外，CIFS採用NT網域協議與認 證，而NT網域協議幾乎只能透過CIFS進行IPS認證，確定客戶端能否拜訪特定資源，所以CIFS所以才會如此廣泛地被採用。
在昇陽電 腦(Sun)於1996年推出WebNFS時，微軟曾經提出將SMB改稱為現在的CIFS，並加入如軟連結、硬連結等功能，藉此提高文件大小。然而，微軟 後來又在推出Vista作業系統時，提出了SMB 2.0的概念，因此目前CIFS與SMB，都是非常普遍的說法。
網路文件共享系統(Common Internet File System；CIFS)原先被稱為伺服器訊息區塊(Server Message Block；SMB)，該技術使得網路上的機器得以共享文件、周邊機器(如印表機)、串接阜等資源，也提供通過授權的程序間溝通(Inter- Process Communication；IPC)功能。
有別於網路文件系統(Network File System；NFS)，CIFS的訊息格式並沒有固定長度，且多數訊息都包含無固定長度的數據，也使得協議的複雜度增加。一般來說，CIFS會包括1個32位元組的訊息頭，以及1個或多個長度不固定的訊息體。
CIFS 最初是由IBM的Barry Feigenbaum所開發，目的是為了將DOS上的本地端文件介面，改造成為可供分享的網路文件系統。後來，微軟(Microsoft)接手了該計畫， 為了修正第1版的DES錯誤，而使用了NTLM(NT Lan Manager)第2版，並將CIFS與微軟和3com共同合作的另1個網路管理程序結合，修改為當前的版本，並建立其分散式文件系統。也因為這樣的緣 故，CIFS多半是用在安裝微軟作業系統的機器上，但後來CIFS在經過Unix伺服器廠商重新開發後，也可以被用在連接Unix伺服器和Windows 作業系統的客戶端機器上，並執行檔案分享，以及共用PC周邊的功能。
CIFS最早是被設計成運行於NetBIOS協議上；換言之，CIFS是建立在NetBEUI、IPX/SPX以及TCP/IP協議成立的前提之上。後來，微軟在Windows 2000作業系統中，將CIFS改造為可以直接運行在TCP/IP之上。
CIFS 的作業方式，基本上就是大家熟悉的客戶端－伺服器(Client - Server)模式，也就是先由客戶端發出請求，再由伺服器回應；在CIFS協定中，有一部分就是用來協調客戶端對伺服器發出的請求，也就是說，客戶端是 藉由文件系統的中介，得以拜訪文件伺服器。在IPC部分，CIFS也允許利用網際網路功能，讓不同的局部子網路或虛擬私人網路(Virtual Private Network；VPN)得以互相連結，如此一來，不同的子網路之間就得以進行遠端協同作業，例如文件分享或是共用印表機等PC周邊設備功能，都是利用 CIFS協議運作。
CIFS最為人詬病的缺點，就是它會佔用過多頻寬，因為每1台子網路內的客戶端電腦，都必須要廣播其存在，才能共享服 務，但是CIFS本身並不使用廣播功能，這樣的問題實際上是NetBIOS的服務定位協議所造成的。在使用微軟Windows伺服器時，都是透過 NetBIOS來協議與定位，而NetBIOS則會定時向特定伺服器廣播某個服務的存在，這樣的作業方式在伺服器較少時，並不會佔用過多的頻寬，但在伺服 器超過一定數目時，網路大塞車的情況則會愈來愈明顯。
網路塞車會對CIFS的使用者帶來重大的影響，特別是在伺服器間的網路延遲時間過長時，透過網際網路拜訪VPN，進行文件分享與協同作業時，延遲會更加明顯。
為 了改善頻寬問題，一般來說可以利用Windows Internet Name Service(WINS)定位協議緩和這樣的情況，也就是使用更高階的系統將服務集中化，但是這也相對地會造成訊息與網路保管問題。此外，使用動態網域 名稱伺服器(Dynamic Domain Name Server；Dynamic DNS)也可以解決CIFS的問題。
CIFS伺服 器可以向網路上的客戶端電腦提供文件系統與各種資源，但客戶端卻可以保留分享權限，這使得CIFS得到廣泛的採用。此外，CIFS採用NT網域協議與認 證，而NT網域協議幾乎只能透過CIFS進行IPS認證，確定客戶端能否拜訪特定資源，所以CIFS所以才會如此廣泛地被採用。
在昇陽電 腦(Sun)於1996年推出WebNFS時，微軟曾經提出將SMB改稱為現在的CIFS，並加入如軟連結、硬連結等功能，藉此提高文件大小。然而，微軟 後來又在推出Vista作業系統時，提出了SMB 2.0的概念，因此目前CIFS與SMB，都是非常普遍的說法。
DIGITIMES中文網 原文網址: 網路文件共享系統(CIFS) http://www.digitimes.com.tw/tw/dt/n/shwnws.asp?rnrnCnlID=10&Cat=&Cat1=&id=124179#ixzz29bnGeE00
Server Message Block
The Server Message Block protocol can run atop the Session (and lower) network layers in several ways:
- directly over TCP, port 445;[2]
- via the NetBIOS API, which in turn can run on several transports:[3]
- on UDP ports 137, 138 & TCP ports 137, 139 — see NetBIOS over TCP/IP;
- on several legacy protocols such as NBF (incorrectly referred to as NetBEUI).
History
Barry Feigenbaum originally designed SMB at IBM with the aim of turning DOS "Interrupt 33" (21h) local file access into a networked file system.[4] Microsoft has made considerable modifications to the most commonly used version. Microsoft merged the SMB protocol with the LAN Manager product which it had started developing for OS/2 with 3Com c. 1990, and continued to add features to the protocol in Windows for Workgroups (c. 1992) and in later versions of Windows.SMB was originally designed to run on top of the NetBIOS/NetBEUI API (typically implemented with NBF, NetBIOS over IPX/SPX, or NBT). Since Windows 2000, SMB runs, by default, with a thin layer, similar to the Session Message packet of NBT's Session Service, on top of TCP, using TCP port 445 rather than TCP port 139 — a feature known as "direct host SMB".[2]
At around the time when Sun Microsystems announced WebNFS,[5] Microsoft launched an initiative in 1996 to rename SMB to Common Internet File System (CIFS), and added more features, including support for symbolic links, hard links, larger file sizes, and an initial attempt at supporting direct connections over TCP port 445 without requiring NetBIOS as a transport (a largely experimental effort that required further refinement). Microsoft submitted some partial specifications as Internet-Drafts to the IETF,[6] though these submissions have expired.
The Samba project originated with the aim of reverse engineering the SMB protocol and implementing an SMB server to allow MS-DOS clients to use SMB to access files on Sun Microsystems machines.[7] Because of the importance of the SMB protocol in interacting with the widespread Microsoft Windows platform, Samba became a popular free implementation of a compatible SMB client and server for interoperating with non-Microsoft operating systems.
Microsoft introduced SMB2 with Windows Vista in 2006, and later improved on it in Windows 7.
Implementation
Client-server approach
SMB works through a client-server approach, where a client makes specific requests and the server responds accordingly. One section of the SMB protocol specifically deals with access to filesystems, such that clients may make requests to a file server; but some other sections of the SMB protocol specialize in inter-process communication (IPC). The Inter-Process Communication (IPC) share, or ipc$, is a network share on computers running Microsoft Windows. This virtual share is used to facilitate communication between processes and computers over SMB, often to exchange data between computers that have been authenticated.Developers have optimized the SMB protocol for local subnet usage, but users have also put SMB to work to access different subnets across the Internet—exploits involving file-sharing or print-sharing in MS Windows environments usually focus on such usage.
SMB servers make their file systems and other resources available to clients on the network. Client computers may want access to the shared file systems and printers on the server, and in this primary functionality SMB has become best-known and most heavily used. However, the SMB file-server aspect would count for little without the NT domains suite of protocols, which provide NT-style domain-based authentication at the very least. Almost all implementations of SMB servers use NT Domain authentication to validate user-access to resources.
Samba
Main article: Samba (software)
Samba is a free software re-implementation of the SMB/CIFS networking protocol, originally developed by Andrew Tridgell. As of version 3, Samba provides file and print services for Microsoft Windows clients and can integrate with a Windows NT 4.0 server domain, either as a Primary Domain Controller (PDC) or as a domain member. Samba4 installations can act as an Active Directory domain controller or member server, at Windows 2008 domain and forest functional levels.[8]Performance issues
NetBIOS
The use of the SMB protocol has often correlated with a significant increase in broadcast traffic on a network. However the SMB itself does not use broadcasts—the broadcast problems commonly associated with SMB actually originate with the NetBIOS service location protocol. By default, a Microsoft Windows NT 4.0 server used NetBIOS to advertise and locate services. NetBIOS functions by broadcasting services available on a particular host at regular intervals. While this usually makes for an acceptable default in a network with a smaller number hosts, increased broadcast traffic can cause problems as the size of the network increases. The implementation of name resolution infrastructure in the form of Windows Internet Naming Service (WINS) or Domain Name System (DNS) resolves this problem. WINS was a proprietary implementation used with Windows NT 4.0 networks, but brought about its own issues and complexities in the design and maintenance of a Microsoft network.Since the release of Windows 2000, the use of WINS for name resolution has been deprecated by Microsoft, with hierarchical Dynamic DNS now configured as the default name resolution protocol for all Windows operating systems. Resolution of (short) NETBIOS names by DNS requires that a DNS client expand short names, usually by appending a connection-specific DNS suffix to its DNS lookup queries. WINS can still be configured on clients as a secondary name resolution protocol for interoperability with legacy Windows environments and applications. Further, Microsoft DNS servers can forward name resolution requests to legacy WINS servers in order to support name resolution integration with legacy (pre-Windows 2000) environments that do not support DNS.
WAN performance issues
Network designers have found that latency has a significant impact on the performance of the SMB 1.0 protocol, that it performs more poorly than other protocols like FTP. Monitoring reveals a high degree of "chattiness" and a disregard of network latency between hosts.[9] For example, a VPN connection over the Internet will often introduce network latency. Microsoft has explained that performance issues come about primarily because SMB 1.0 is a block-level rather than a streaming protocol, that was originally designed for small LANs; it has a block size that is limited to 64K, SMB signing creates an additional overhead and the TCP window size is not optimized for WAN links.[10] Solutions to this problem include the updated SMB 2.0 protocol, Offline Files, TCP window scaling and WAN acceleration devices from various network vendors that cache and optimize SMB 1.0[11] and 2.0.[12]Microsoft's modifications
Microsoft added several extensions to its own SMB implementation. For example, it added NTLM, then NTLMv2 authentication protocols in order to address security weakness in the original LanMan authentication. LanMan authentication derived from the original legacy SMB specification's requirement to use IBM "LanManager" passwords, but implemented DES in a flawed manner that allowed passwords to be cracked.[13] Later, Kerberos authentication was also added. The NT 4.0 Domain logon protocols initially used 40-bit encryption outside of the United States of America, because of export restrictions on stronger 128-bit encryption[14] (subsequently lifted in 1996 when President Bill Clinton signed Executive Order 13026[15]). Opportunistic locking support has changed with each server release.Opportunistic locking
 |
This section does not cite any references or sources. (February 2011) |
Batch Locks
Batch OpLocks were created originally to support a particular behavior of MS-DOS batch file execution operation in which the file is opened and closed many times in a short period, which is a performance problem. To solve this, a client may ask for an OpLock of type "batch". In this case, the client delays sending the close request and if a subsequent open request is given, the two requests cancel each other.Exclusive Locks
When an application opens in "shared mode" a file hosted on an SMB server which is not opened by any other process (or other clients) the client receives an exclusive OpLock from the server. This means that the client may now assume that it is the only process with access to this particular file, and the client may now cache all changes to the file before committing it to the server. This is a performance improvement, since fewer round-trips are required in order to read and write to the file. If another client/process tries to open the same file, the server sends a message to the client (called a break or revocation) which invalidates the exclusive lock previously given to the client. The client then flushes all changes to the file.Level 2 OpLocks
If an exclusive OpLock is held by a client and a locked file is opened by a third party, the client has to relinquish its exclusive OpLock to allow the other client's write/read access. A client may then receive a "Level 2 OpLock" from the server. A Level 2 OpLock allows the caching of read requests, but excludes write caching.Breaks
In contrast with the SMB protocol's "standard" behavior, a break request may be sent from server to client. It informs the client that an OpLock is no longer valid. This happens, for example, when another client wishes to open a file in a way that invalidates the OpLock. The first client is then sent an OpLock break and required to send all its local changes (in case of batch or exclusive OpLocks), if any, and acknowledge the OpLock break. Upon this acknowledgment the server can reply to the second client in a consistent manner.SMB2
Microsoft introduced a new version of the Server Message Block (SMB) protocol (SMB 2.0 or SMB2) with Windows Vista in 2006.[16] Although the protocol is proprietary, its specification has been published to allow other systems to interoperate with Microsoft operating systems that use the new protocol.[17]SMB2 reduces the 'chattiness' of the SMB 1.0 protocol by reducing the number of commands and subcommands from over a hundred to just nineteen.[9] It has mechanisms for pipelining, that is, sending additional requests before the response to a previous request arrives, thereby improving performance over high latency links. It adds the ability to compound multiple actions into a single request, which significantly reduces the number of round-trips the client needs to make to the server, improving performance as a result.[9] SMB1 also has a compounding mechanism — known as AndX — to compound multiple actions, but Microsoft clients rarely use AndX.[citation needed] It also introduces the notion of "durable file handles": these allow a connection to an SMB server to survive brief network outages, as are typical in a wireless network, without having to incur the overhead of re-negotiating a new session.
SMB2 includes support for symbolic links. Other improvements include caching of file properties, improved message signing with HMAC SHA-256 hashing algorithm and better scalability by increasing the number of users, shares and open files per server among others.[9] The SMB1 protocol uses 16-bit data sizes, which amongst other things, limits the maximum block size to 64K. SMB2 uses 32 or 64-bit wide storage fields, and 128 bits in the case of file-handles, thereby removing previous constraints on block sizes, which improves performance with large file transfers over fast networks.[9]
Windows Vista/Server 2008 and later operating systems use SMB2 when communicating with other machines also capable of using SMB2. SMB1 continues in use for connections with older versions of Windows, as well as systems like Samba and various vendors' NAS solutions. Samba 3.5 also includes experimental support for SMB2.[18] Samba 3.6 fully supports SMB2, except the modification of user quotas using the Windows quota management tools.[19]
When SMB2 was introduced it brought a number of benefits over SMB1 for third party implementers of SMB protocols. SMB1, originally designed by IBM, was reverse engineered, and later became part of a wide variety of non-Windows operating systems such as Xenix, OS/2 and VMS (Pathworks). X/Open standardised it partially; it also had draft IETF standards which lapsed. (See http://ubiqx.org/cifs/Intro.html for historical detail.) SMB2 is also a relatively clean break with the past. Microsoft's SMB1 code has to work with a large variety of SMB clients and servers. SMB1 features many versions of information for commands (selecting what structure to return for a particular request) because features such as Unicode support were retro-fitted at a later date. SMB2 involves significantly reduced compatibility-testing for implementers of the protocol. SMB2 code has considerably less complexity since far less variability exists (for example, non-Unicode code paths become redundant as SMB2 requires Unicode support).
SMB 2.1
SMB 2.1, introduced with Windows 7 and Server 2008 R2, introduced minor performance enhancements with a new opportunistic locking mechanism.[20]SMB 3.0
SMB 3.0 (previously named SMB 2.2)[21] was introduced with Windows 8[21] and Windows Server 2012.[21] It will bring several significant changes that are aimed to add functionality and improve SMB2 performance, notably in virtualized data centers like SMB2 RDMA Transport Protocol and multichannel.[22]Features
The SMB "Inter-Process Communication" (IPC) system provides named pipes and was one of the first inter-process mechanisms commonly available to programmers that provides a means for services to inherit the authentication carried out when a client[clarification needed] first connected to an SMB server.[citation needed]Some services that operate over named pipes, such as those which use Microsoft's own implementation of DCE/RPC over SMB, known as MSRPC over SMB, also allow MSRPC client programs to perform authentication, which over-rides the authorization provided by the SMB server, but only in the context of the MSRPC client program that successfully makes the additional authentication.
Since Windows domain controllers use SMB to transmit policies at login, they have packet-signing enabled by default to prevent man-in-the-middle attacks; the feature can also be turned on for any server running Windows NT 4.0 Service Pack 3 or later.[23] The design of Server Message Block version 2 (SMB2) aims[citation needed] to mitigate this performance-limitation by coalescing SMB signals into single packets.
SMB supports opportunistic locking — a special type of locking-mechanism — on files in order to improve performance.
SMB serves as the basis for Microsoft's Distributed File System implementation.
Security
Over the years, there have been many security vulnerabilities in Microsoft's implementation of the protocol or components that it directly relies on,[24][25][26] with the most recent vulnerability (at time of writing) involving the SMB2 implementation.[27] Other vendors' security vulnerabilities lie primary in a lack of support for newer authentication protocols like NTLMv2 and Kerberos in preference to broken protocols like NTLMv1, LanMan, or even plaintext passwords.Specifications for SMB and SMB2 Protocols
The specifications for the SMB are proprietary and were originally closed, thereby forcing other vendors and projects to reverse-engineer the protocol in order to interoperate with it. The SMB 1.0 protocol was eventually published some time after it was reverse engineered, whereas the SMB 2.0 procotol was made available from Microsoft's MSDN Open Specifications Developer Center from the outset.[28] There are a number of specifications that are relevant to the SMB protocol:- MS-CIFS [1] MS-CIFS is a recent replacement (2007) for the draft-leach-cifs-v1-spec-02.txt a document widely used to implement SMB clients, but also known to have errors of omission and commission.
- MS-SMB [2] Specification for Microsoft Extensions to MS-CIFS
- MS-SMB2 [3] Specification for the SMB 2 protocol
- MS-FSSO [4] Describes the intended functionality of the Windows File Access Services System, how it interacts with systems and applications that need file services, and how it interacts with administrative clients to configure and manage the system.
沒有留言:
張貼留言