Tuxedo

Tuxedo (Transactions for Unix, Extended for Distributed Operations) is a middleware platform used to manage distributed transaction processing in distributed computing environments. Tuxedo is a transaction processing system or transaction oriented middleware, or enterprise application server for C, C++, and COBOL.

It was originally developed and designed by AT&T in 1983 for the creation and administration of operations support systems that required online transaction processing (OLTP) capabilities. The Tuxedo concepts were derived from the LMOS system. The original Tuxedo team was composed of seasoned members of the LMOS team. In 1993 Novell acquired the Unix System Laboratories (USL) division of AT&T which was responsible for the development of Tuxedo at the time. In 1996, BEA Systems made an exclusive agreement with Novell to develop and distribute Tuxedo on non-NetWare platforms, with most Novell employees working with Tuxedo joining BEA. In 2008, Oracle Corporation acquired BEA Systems.^[1][2]

Tuxedo was designed from the beginning for high availability and to provide extremely scalable applications allowing Tuxedo to support applications requiring thousands of transactions per second on commonly available distributed systems. One of the first applications within AT&T for Tuxedo was to support moving the LMOS application off mainframe systems on to much cheaper distributed systems.

Tuxedo has been used as transactional middleware by a number of multi-tier application development tools.

Features of Tuxedo

• Standards based APIs - SCA, The Open Group XATMI, Object Management Group CORBA
• Communication types - Synchronous, Asynchronous, Conversational, Unsolicited Notifications, Publish/subscribe
• Typed buffers
  • FML/FML32 - Self describing fielded buffers similar to Abstract Syntax Notation One or Fast Infoset
  • XML
  • STRING and multbyte strings MBSTRING
  • CARRAY binary blobs
  • VIEW/VIEW32 externally descripted records
• Transaction Management - Global Transactions - Two-phase commit protocol - X/Open XA
• Clustering - Domains
• /WS - Remote Clients
• Java clients - Jolt
• Bidirectional Web Services - SALT
• /QUEUE - Transient (in memory) and Persistent Queues (also called Reliable Queues)
• Data Dependent Routing (DDR)
• Event Broker
• Security - Authentication, Authorization, Auditing, and Public key infrastructure based message signing and encryption
• Programmed Administration and SNMP support
• System and application performance monitoring - TSAM
• Load balancing, server spawning and decay
• Supports C, C++, and COBOL applications on most Unix platforms, Linux, Microsoft Windows, and other proprietary platforms

Messaging core

Tuxedo is at its core a message routing and queuing system. Requests are sent to named services and Tuxedo uses memory based inter-process communication facilities to queue the requests to servers. The requester is unaware of where the server that actually processes the request is located or how it is implemented. In essence, Tuxedo provided the elements of Service-oriented architecture (SOA) decades before the phrase Service-oriented architecture was coined. Tuxedo can use the content of the message to determine what servers should be utilized to receive the request by means of data dependent routing.

Clustering

The heart of the Tuxedo system is the Bulletin Board (BB). This is a shared memory segment that contains the state of a Tuxedo domain. Servers, services, transactions, and clients are all registered in the BB providing a global view of their state across the machines within a domain. To coordinate updates to the BB a process called the Bulletin Board Liaison (BBL) runs on each machine to keep the local copy of the BB up-to-date. A master machine runs a process called the “Distinguished Bulletin Board Liaison” that coordinates the updates to the BB. This allows each machine to have a view of what servers, services, transactions, and clients are on each machine within the domain.

Another process on each machine called the Bridge is responsible for passing requests from one machine to another. This allows Tuxedo to spread load across the various machines within a domain and allows servers and services to be running on multiple machines. In addition the BBL and Bridge monitor each other and restart the other should one fail. In the advent of a failure of the master machine, another machine designated as a backup master can take over the function of master machine. Also, since machines within a single domain can be of different architectures (x86, IA32, SPARC, P-Series, etc.), the Bridge is also responsible for handling differences in things like endianness.

Flexible buffer formats

Tuxedo applications can utilize a variety of message formats depending upon the type of data that is to be passed. One of the most popular formats is the FML buffer format which is much like a binary XML or ASN.1 format. FML buffers can contain an arbitrary number of named fields of arbitrary type. Fields can be repeated and nested. As it is a self describing binary format, the processing of fields incurs very little overhead in comparison to the parsing necessary to support something like XML. VIEW buffers are essentially records, C structures, or COBOL copybooks. A VIEW buffer has an external description which allows Tuxedo to access the fields within it if necessary for things like data dependent routing. Other buffer formats include XML, CARRAY (opaque binary data), STRING, and MBSTRING (a string buffer containing multibyte characters.) Tuxedo can automatically and transparently convert FML buffers to and from XML buffers.

There is also support for user-developed buffer types (for example JamFlex buffers defined by Tuxedo version of Panther RAD toolset).

Communication concentrators

For remote clients (Java, CORBA, or /WS), Tuxedo provides communication concentrators called listener/handlers that handle the remote network communication. Clients connect to these communication concentrators and act as proxies for the clients. As clients make requests, the listener/handler uses the local Tuxedo infrastructure to make the request on the behalf of the client. Tuxedo then load balances the requests across the servers within the domain that offer the service even if the server is not on the local machine. This is in contrast to most Java EE application servers where load balancing is done by the client making requests to different machines with the cluster.

Gateways

To facilitate the sharing of services across domains, Tuxedo provides domain gateways. A domain gateway's job is to allow the importing and exporting of services from remote domains. This allows the local domain to see services on remote domains as though they were local services. The domain gateways are responsible for propagating security and transaction context to the remote domain. Besides connecting Tuxedo domains together, domain gateways exist for mainframe systems using TCP/IP, SNA, or the OSI/TP protocols, and Java EE application servers. For the mainframe gateways, each system sees the services imported from the remote system as local services and use the local systems infrastructure to interact with those services. This means that Tuxedo sees a CICS transaction as a Tuxedo service, and CICS sees a Tuxedo service as a CICS transaction.

Failure recovery

The BBL on each machine monitors the state of all servers and can automatically restart failed servers. It can also detect hung servers and kill/restart them as required. Any transactions that are affected by a server or machine failure are rolled back.

Transaction monitoring and coordination

Tuxedo applications can request that all service invocations and their associated updates to any resources controlled by resource managers (such as databases) be controlled by a transaction. Once the application begins a transaction, all subsequent service invocations and nested invocations are included as part of that transaction, even those services that were executed on remote domains. Tuxedo then coordinates the commit processing with the resource managers to ensure atomic updates to all affected resources. Transactions can be controlled by the application or automatically controlled by the Tuxedo configuration, i.e., container controlled transactions.

Queuing subsystem

Tuxedo provides a queuing subsystem called /Q. This facility provides transient and persistent queues that allows application to explicitly queue requests to named queues. Queues can be ordered by message availability time, expiration time, priority, LIFO, FIFO, or a combination. Queues are managed by an XA compliant resource manager allowing queue operations to participate in transactions. An automated queue forwarding server is provided that will remove entries from a queue and invoke an associated Tuxedo services.

Event subsystem

The event subsystem within Tuxedo provides support for unsolicited events as well as brokered events. Unsolicited events allow Tuxedo applications to send out-of-band notifications to clients that aren't necessarily waiting for a response. Brokered events allow application to subscribe to events of interest and when another application posts an event, all applications subscribed to that event receive it. This allows applications to use an event driven model instead of the more typical request/response model.

References

External links

• BEA Tuxedo website
• Tuxedo Monitor Tools