I am writing a service which serves stateless incoming requests. The requests are all mathematical calculation, which does not take very long to execute ( max 2ms).
I use Tibco EMS to communicate between client/server. A client library is provided which wraps the client side logic (e.g. convert data into EMS message etc) and send the request to a request queue. The server side processes the request and send the response into a separate queue. This works fine.
The server side is multi-threaded. A new thread is created when a new incoming request is received. Requests are therefore handled concurrently.
The server side uses one single EMS connection to the EMS server. However, because EMS Session is not thread safe, if I want to be able to write the response to EMS queue in each thread, I have to create one session for each thread using the connectionFactory. This degraded the performance.
The time spent on traffic is around 3-4ms, i.e. Time between a request is sent and a response is received is around 5-6ms.(3-4ms for transportation, marshal/unmarshal, 2ms for calculation).
Is there any solution which allows me to concurrently send to a EMS queue without creating two much JMS objects?
Are there any other important rules I need to follow to further optimize the service? Some basic optimization guidelines are already followed:
Use CachedConnectionPool
Send JMS message as NON_PERSISTANT
Use one EMS connection for all requests.
Thank you very much.
The behavior you are experiencing is not specific to EMS. The behavior is dictated by the JMS specification itself. Here is an extract from section 2.8 of the JMS Specification:
There are two reasons for restricting concurrent access to Sessions. First, Sessions are the JMS entity that supports transactions. It is very difficult to implement transactions that are multithreaded. Second, Sessions support asynchronous message consumption. It is important that JMS not require that client code used for asynchronous message consumption be capable of handling multiple, concurrent messages. In addition, if a Session has been set up with multiple, asynchronous consumers, it is important that the client is not forced to handle the case where these separate consumers are concurrently executing. These restrictions make JMS easier to use for typical clients. More sophisticated clients can get the concurrency they desire by using multiple sessions.
If you want to avoid the creation (and destruction) of that many objects, you might want to pre-create a pool of threads, and allocate a session to each thread up front.
The RabbitMQ Java client has the following concepts:
Connection - a connection to a RabbitMQ server instance
Channel - ???
Consumer thread pool - a pool of threads that consume messages off the RabbitMQ server queues
Queue - a structure that holds messages in FIFO order
I'm trying to understand the relationship, and more importantly, the associations between them.
I'm still not quite sure what a Channel is, other than the fact that this is the structure that you publish and consume from, and that it is created from an open connection. If someone could explain to me what the "Channel" represents, it might help clear a few things up.
What is the relationship between Channel and Queue? Can the same Channel be used to communicate to multiples Queues, or does it have to be 1:1?
What is the relationship between Queue and the Consumer Pool? Can multiple Consumers be subscribed to the same Queue? Can multiple Queues be consumed by the same Consumer? Or is the relationship 1:1?
A Connection represents a real TCP connection to the message broker, whereas a Channel is a virtual connection (AMQP connection) inside it. This way you can use as many (virtual) connections as you want inside your application without overloading the broker with TCP connections.
You can use one Channel for everything. However, if you have multiple threads, it's suggested to use a different Channel for each thread.
Channel thread-safety in Java Client API Guide:
Channel instances are safe for use by multiple threads. Requests into
a Channel are serialized, with only one thread being able to run a
command on the Channel at a time. Even so, applications should prefer
using a Channel per thread instead of sharing the same Channel across
multiple threads.
There is no direct relation between Channel and Queue. A Channel is used to send AMQP commands to the broker. This can be the creation of a queue or similar, but these concepts are not tied together.
Each Consumer runs in its own thread allocated from the consumer thread pool. If multiple Consumers are subscribed to the same Queue, the broker uses round-robin to distribute the messages between them equally. See Tutorial two: "Work Queues".
It is also possible to attach the same Consumer to multiple Queues.
You can understand Consumers as callbacks. These are called everytime a message arrives on a Queue the Consumer is bound to. For the case of the Java Client, each Consumers has a method handleDelivery(...), which represents the callback method. What you typically do is, subclass DefaultConsumer and override handleDelivery(...). Note: If you attach the same Consumer instance to multiple queues, this method will be called by different threads. So take care of synchronization if necessary.
A good conceptual understanding of what the AMQP protocol does "under the hood" is useful here. I would offer that the documentation and API that AMQP 0.9.1 chose to deploy makes this particularly confusing, so the question itself is one which many people have to wrestle with.
TL;DR
A connection is the physical negotiated TCP socket with the AMQP server. Properly-implemented clients will have one of these per application, thread-safe, sharable among threads.
A channel is a single application session on the connection. A thread will have one or more of these sessions. AMQP architecture 0.9.1 is that these are not to be shared among threads, and should be closed/destroyed when the thread that created it is finished with it. They are also closed by the server when various protocol violations occur.
A consumer is a virtual construct that represents the presence of a "mailbox" on a particular channel. The use of a consumer tells the broker to push messages from a particular queue to that channel endpoint.
Connection Facts
First, as others have correctly pointed out, a connection is the object that represents the actual TCP connection to the server. Connections are specified at the protocol level in AMQP, and all communication with the broker happens over one or more connections.
Since it's an actual TCP connection, it has an IP Address and Port #.
Protocol parameters are negotiated on a per-client basis as part of setting up the connection (a process known as the handshake.
It is designed to be long-lived; there are few cases where connection closure is part of the protocol design.
From an OSI perspective, it probably resides somewhere around Layer 6
Heartbeats can be set up to monitor the connection status, as TCP does not contain anything in and of itself to do this.
It is best to have a dedicated thread manage reads and writes to the underlying TCP socket. Most, if not all, RabbitMQ clients do this. In that regard, they are generally thread-safe.
Relatively speaking, connections are "expensive" to create (due to the handshake), but practically speaking, this really doesn't matter. Most processes really will only need one connection object. But, you can maintain connections in a pool, if you find you need more throughput than a single thread/socket can provide (unlikely with current computing technology).
Channel Facts
A Channel is the application session that is opened for each piece of your app to communicate with the RabbitMQ broker. It operates over a single connection, and represents a session with the broker.
As it represents a logical part of application logic, each channel usually exists on its own thread.
Typically, all channels opened by your app will share a single connection (they are lightweight sessions that operate on top of the connection). Connections are thread-safe, so this is OK.
Most AMQP operations take place over channels.
From an OSI Layer perspective, channels are probably around Layer 7.
Channels are designed to be transient; part of the design of AMQP is that the channel is typically closed in response to an error (e.g. re-declaring a queue with different parameters before deleting the existing queue).
Since they are transient, channels should not be pooled by your app.
The server uses an integer to identify a channel. When the thread managing the connection receives a packet for a particular channel, it uses this number to tell the broker which channel/session the packet belongs to.
Channels are not generally thread-safe as it would make no sense to share them among threads. If you have another thread that needs to use the broker, a new channel is needed.
Consumer Facts
A consumer is an object defined by the AMQP protocol. It is neither a channel nor a connection, instead being something that your particular application uses as a "mailbox" of sorts to drop messages.
"Creating a consumer" means that you tell the broker (using a channel via a connection) that you would like messages pushed to you over that channel. In response, the broker will register that you have a consumer on the channel and begin pushing messages to you.
Each message pushed over the connection will reference both a channel number and a consumer number. In that way, the connection-managing thread (in this case, within the Java API) knows what to do with the message; then, the channel-handling thread also knows what to do with the message.
Consumer implementation has the widest amount of variation, because it's literally application-specific. In my implementation, I chose to spin off a task each time a message arrived via the consumer; thus, I had a thread managing the connection, a thread managing the channel (and by extension, the consumer), and one or more task threads for each message delivered via the consumer.
Closing a connection closes all channels on the connection. Closing a channel closes all consumers on the channel. It is also possible to cancel a consumer (without closing the channel). There are various cases when it makes sense to do any of the three things.
Typically, the implementation of a consumer in an AMQP client will allocate one dedicated channel to the consumer to avoid conflicts with the activities of other threads or code (including publishing).
In terms of what you mean by consumer thread pool, I suspect that Java client is doing something similar to what I programmed my client to do (mine was based off the .Net client, but heavily modified).
I found this article which explains all aspects of the AMQP model, of which, channel is one. I found it very helpful in rounding out my understanding
https://www.rabbitmq.com/tutorials/amqp-concepts.html
Some applications need multiple connections to an AMQP broker. However, it is undesirable to keep many TCP connections open at the same time because doing so consumes system resources and makes it more difficult to configure firewalls. AMQP 0-9-1 connections are multiplexed with channels that can be thought of as "lightweight connections that share a single TCP connection".
For applications that use multiple threads/processes for processing, it is very common to open a new channel per thread/process and not share channels between them.
Communication on a particular channel is completely separate from communication on another channel, therefore every AMQP method also carries a channel number that clients use to figure out which channel the method is for (and thus, which event handler needs to be invoked, for example).
There is a relation between like A TCP connection can have multiple Channels.
Channel: It is a virtual connection inside a connection. When publishing or consuming messages from a queue - it's all done over a channel Whereas Connection: It is a TCP connection between your application and the RabbitMQ broker.
In multi-threading architecture, you may need a separate connection per thread. That may lead to underutilization of TCP connection, also it adds overhead to the operating system to establish as many TCP connections it requires during the peak time of the network. The performance of the system could be drastically reduced. This is where the channel comes handy, it creates virtual connections inside a TCP connection. It straightaway reduces the overhead of the OS, also it allows us to perform asynchronous operations in a more faster, reliable and simultaneously way.
I'm learning JMS and wonder how a JMS client (e.g MessageListener) can notice about a new message in queue it registed. Is it frequently send requests to broker via TCP to see if there's a new message? If so, is this request synchronousor asynchronous?
JMS is just an API. It does not specify any wire level protocol. So you can't really tell how the client will behave with the broker. It could use a homing piegon for all we know. Ok, maybe not, but brokers like WebSphere MQ and ActiveMQ both supply in memory transport as well as TCP based.
Most vendors have thier own properitary protocols even though AMQP is visible on the horizon as a wire protocol standard (but far from all vendors have started to look at it).
When talking TCP there is no need to poll as long as there is a live connection going on. The broker can easily notify the client that there is a new message published while the client sleeps and the other way around.
A common way, however, is to actually do poll. But rather poll for consumer.receive(TIMEOUT); in some longer intervals (seconds). This makes it possible to use distributed transactions in frameworks like spring. Still the broker sends actual TCP messages to the client on demand.
If it would not have been like this, then JMS/Messaging would not have been such a fast, wide psread and scalable technology
1) First of all, JMS does not have something called absolute synchronous messaging. You can definitely implement so called JMS Synchronous messaging by implementing Sync service methods but in fact it just appears to be mimicking as Synchronous messaging. In fact it is also Async Messaging.
2) Technically it is the JMS Server / Broker which sends Messages to Message Consumers through dedicated queues. Broker simply delivers the message to Message Consumer's onMessage() method. And then Container executes onMessage() method.
We are running a high throughput system that utilizes tibco-ems JMS to pass large numbers of messages to and from our main server to our client connections. We've done some statistics and have determined that JMS is the causing a lot of latency. How can we make tibco JMS more performant? Are there any resources that give a good discussion on this topic.
Using non-persistent messages is one option if you don't need persistence.
Note that even if you do need persistence, sometimes it's better to use non persistent messages, and in case of a crash perform a different recovery action (like resending all messages)
This is relevant if:
crashes are rare (as the recovery takes time)
you can easily detect a crash
you can handle duplicate messages (you may not know exactly which messages were delivered before the crash
EMS also provides some mechanisms that are persistent, but less bullet proof then classic guaranteed delivery
these include:
instead of "exactly once" message delivery you can use "at least once" or "up to once" delivery.
you may use the pre-fetch mechanism which causes the client to fetch messages to memory before your application request them.
EMS should not be the bottle neck. I've done testing and we have gotten a shitload of throughput on our server.
You need to try to determine where the bottle neck is. Is the problem in the producer of the message or the consumer. Are messages piling up on the queue.
What type of scenario are you doing.
Pub/sup or request reply?
are you having temporary queue pile up. Too many temporary queues can cause performance issues. (Mostly when they linger because you didn't close something properly)
Are you publishing to a topic with durable subscribers if so. Try bridging the topic to queue and reading from those. Durable subscribers can cause a little hiccup in performance too since it needs to track who has copies of all messages.
Ensure that your sending process has one session and multiple calls through that session. Don't open a complete session for each operation. Re-use where possible. Do the same for the consumer.
make sure you CLOSE when you are done. EMS doesn't clear things up. So if you make a connection and just close your app the connection still is there and sucking up resources.
review your tolerance for lost messages in the even of a crash. If you are doing Client ack and it doesn't matter if you crash processing the message then switch to auto. Also I believe if you are using (TEMS - Tibco EMS for WCF) there's a problem with the session acknowledge. So a message is only when its processed on the whole message, we switched from Client ACK to the one that had Dups ok and it worked better)
My Java EE application sends JMS to queue continuously, but sometimes the JMS consumer application stopped receiving JMS. It causes the JMS queue very large even full, that collapses the server.
My server is JBoss or Websphere. Do the application servers provide strategy to remove "timeout" JMS messages?
What is strategy to handle large JMS queue? Thanks!
With any asynchronous messaging you must deal with the "fast producer/slow consumer" problem. There are a number of ways to deal with this.
Add consumers. With WebSphere MQ you can trigger a queue based on depth. Some shops use this to add new consumer instances as queue depth grows. Then as queue depth begins to decline, the extra consumers die off. In this way, consumers can be made to automatically scale to accommodate changing loads. Other brokers generally have similar functionality.
Make the queue and underlying file system really large. This method attempts to absorb peaks in workload entirely in the queue. This is after all what queuing was designed to do in the first place. Problem is, it doesn't scale well and you must allocate disk that 99% of the time will be almost empty.
Expire old messages. If the messages have an expiry set then you can cause them to be cleaned up. Some JMS brokers will do this automatically while on others you may need to browse the queue in order to cause the expired messages to be deleted. Problem with this is that not all messages lose their business value and become eligible for expiry. Most fire-and-forget messages (audit logs, etc.) fall into this category.
Throttle back the producer. When the queue fills, nothing can put new messages to it. In WebSphere MQ the producing application then receives a return code indicating that the queue is full. If the application distinguishes between fatal and transient errors, it can stop and retry.
The key to successfully implementing any of these is that your system be allowed to provide "soft" errors that the application will respond to. For example, many shops will raise the MAXDEPTH parameter of a queue the first time they get a QFULL condition. If the queue depth exceeds the size of the underlying file system the result is that instead of a "soft" error that impacts a single queue the file system fills and the entire node is affected. You are MUCH better off tuning the system so that the queue hits MAXDEPTH well before the file system fills but then also instrumenting the app or other processes to react to the full queue in some way.
But no matter what else you do, option #4 above is mandatory. No matter how much disk you allocate or how many consumer instances you deploy or how quickly you expire messages there is always a possibility that your consumer(s) won't keep up with message production. When this happens your producer app should throttle back, or raise an alarm and stop or do anything other than hang or die. Asynchronous messaging is only asynchronous up to the point that you run out of space to queue messages. After that your apps are synchronous and must gracefully handle that situation, even if that means to (gracefully) shut own.
Sure!
http://download.oracle.com/docs/cd/E17802_01/products/products/jms/javadoc-102a/index.html
Message#setJMSExpiration(long) does exactly what you want.