I am currently trying to figure something out about a DDS application I wrote.
My writer and reader currently have reliability on, so that if a reader misses a message the writer will republish the message. I am also using the default multi-cast and not using uni-cast for discovery of publishers and subscribers.
According to the multicast protocol I am using only ports 7000 and 7001 need to be open. However when I did a analysis using wireshark I saw that ports 7010 and 7011 (uni-cast) ports are open as well.
After some digging I came across this link and it seems that to use reliability for readers and writers you need to have uni-cast enabled and this why the uni-cast ports are also open and being used.
Does uni-cast have to really be enabled for having messages delivered reliably and if so why is this needed, as well as why can't multicast do this function?
In this case what is happening is that the majority of the traffic will go out via MC. Occasionally, the reliability protocol will send a message that says, in effect, "I have sequence numbers N through M available."
Each reader will (and this is heavily tunable in the different implementations) respond with something (via unicast!) "ok" or "I didn't get x or z".
If only one reader did not get x, it makes no sense to MC a repair sample x, because only one reader needs it. So the writer will unicast it to the squeaky reader.
That's it in a nutshell, I could spent another 10 paragraphs talking about config options and tuning the behavior.
But yeah tl;dr: expected behavior.
Related
I teach students to develop network applications, both clients and servers. At this moment, we have not yet touched existing protocols such as HTTP, SMTP, etc. The students write very simple programs on top of the plain socket API. Currently I check a students' work manually, but I want to automate this task and create an automated test bench for networking applications. The most interesting topics for testing are:
Breaking TCP segments into small parts and delivering them with a noticeable delay. A reason I need such test is that students usually just issue a read/recv call and process the received data without checking that all necessary data was received. TCP doesn't guarantee the message boundaries, so in certain circumstances it is necessary to make several read/recv calls. The problem is that in most simple network applications (for example, in a chat application) messages are small and fit into the single TCP segment, so the issue doesn't appear. My idea is to artificially break messages into several small TCP segments (i.e. several bytes of data) so the problem will appear.
Pausing the data transfer for some time to simulate multiple slow clients and check that the multithreading/async sockets are implemented properly in the students' servers.
Resetting a connection in random moments of time.
I've found several systems which simulate a bad network (dummynet, clumsy, netem). Hovewer, they all work on the IP level of the stack, so OS and it's TCP implementation will compensate the data loss. Such systems are able to solve the task number 2, but they are not able to solve tasks 1 and 3. So I think that I need to develop my own solution, which will act as a TCP proxy. My questions are:
Maybe the are any libraries or applications which can (at least partially) solve the given tasks, so I'll be able to use them as a base for my own solution?
In case there is none any suitable existing software projects, maybe there are any ideas and approaches about how to do this properly?
From WireShark mailing list - Creating and Modifying Packets:
...There's a "Tools" page on the Wireshark Wiki:
http://wiki.wireshark.org/Tools
which has a "Traffic generators" section:
https://wiki.wireshark.org/Tools#Traffic_generators
which lists some tools that might be useful...
The "Traffic generators" chapter also mentions another collection of traffic generators
If you write your own socket code, you can address all 3 tasks.
enable the socket's TCP_NODELAY option (disable the Nagle Algorithm for Send Coalescing) via setsockopt(), then you can send() small fragments of data as you wish, optionally with a delay in between (see #2).
simply put a delay in between your send() calls.
use setsockopt() to adjust the socket's SO_LINGER and SO_DONTLINGER options to control whether closing the socket performs an abortive or graceful closure, then simply close the socket at some random interval after the connection is established.
I am in a process of learning and trying to use the RTP/RTCP protocol. My situation is that there is 1 to n streamers and 1 (or potentially 1 to m if needed) receiver(s), but in a way that the streamers themselves do not know about each other (they cannot directly due to technical reasons, such as different network, limited bandwidth, etc...). So it is more like multiple unicast sessions, but the receiver actually knows about them all, collects data from all of them, it is just the senders do not know about each other .
Now reading about the protocol, it seems to me that huge portion of it is related to sending some feedback, collision detections, and so on. So I have doubts, is RTP is really applicable in this case? Is is already used in this way somewhere?
Seems to me it is still beneficial to collect statistic about data transfer that RTP provides (data sent, loss, times, etc...), it just feels the most of the protocol is sort of left out...
Also I have one additional question, going through the various RTP libraries, they all assume that sender will also open ports for receiving RTP/RTCP data, does RTP forbid use of one way communication? I mean application that would only stream the data, not expecting to receive anything back. The libraries (e.g. ccRTP) seem to assume both way communication only...
RTCP is the protocol that provides statistics. The stream receiver (client) will send stats to the sender (server) via RTCP. I don't believe the client will get any statistic reports from the server.
There's nothing wrong with a single client receiving multiple unicast sessions from various servers.
RTP requires two way communication during the setup process. Once setup is complete and the play cmd is sent, it is mostly one way. The exception are the "keep alive" packets that must be sent to the server periodically (usually every 60 seconds or so) to keep the stream going. The exact timeout value is sent to the client during the setup process.
But if you implement your own RTP, there's nothing stopping you from having the server send the stream continuously without any feedback from the client. Basically it would be implementing an infinite timeout value.
You can read about all the details in the spec: RTP: A Transport Protocol for Real-Time Applications
I have been reading about ZeroMQ more specifically about NetMQ and almost every Pub/Sub examples I saw used to Bind the Publisher socket and then the Subscriber socket connects to the other.
So i'm wondering if it is possible to do the reverse, i mean Bind the Subscriber socket and then publishers connect to it.
Is this possible ? (I didn't found anything clear on documentation)
What are the disadvantages using this connection strategy ?
Any help will be usefull.
Yes, you can reverse it and there are no disadvantages to using that connection strategy... provided it suits your purpose.
In ZMQ, the driving concept behind "binding" and "connecting" is that one side is often considered to be more reliable (and usually there will be fewer nodes), and the other side is considered to be more transient (and there could be more numerous nodes). The reliable side would be considered your "server", and you should bind() on that side, the transient side would be considered your "client" (or clients), and you should connect() on that side.
Typically, we think of a stable "server" publishing information constantly, to many "client" subscribers which may come and go. This is represented in the examples that you see: bind on pub, connect on sub.
But, you could just as easily have a stable "server" subscribing to any output from many "client" publishers that connect to it, accepting any information that they're sending while they are available. Bind on sub, connect on pub.
You're not limited to one server, either, it's just the simplest example - however, you're more limited if you're running all of your sockets on the same computer. Binding on the same address with more than one socket will produce a conflict, but you can connect as many sockets to the same address as you like.
In many cases, both sides of the communication are really intended to be reliable and long running, in which case it's useful to think of the node which sends the information as the server, and the one which receives it as the client. In which case, we're back to bind on pub, connect on sub.
I have to develop a message bus for processes to send, receive messages from each other. Currently, we are running on Linux with the view of porting to other platforms later.
For this, I am using ZeroMQ over TCP. The pattern is PUB-SUB with a forwarder. My bus runs as a separate process and all clients connect to SUB port to receive messages and PUB to send messages. Each process subscribes to messages by a unique tag. A send call from a process sends messages to all. A receive call will fetch that process the messages marked with the tag of that process. This is working fine.
Now I need to wrap the ZeroMQ stuff. My clients only need to supply a unique tag. I need to maintain a global list of tags vs. ZeroMQ context and sockets details. When a client say,
initialize_comms("name"); the bus needs to check if this name is unique, create ZeroMQ contexts and sockets. Similarly, if a client say receive("name"); the bus needs to fetch messages with that tag.
To summarize the problems I am facing;
Is there anyway to achieve this using facilities provided by ZeroMQ?
Is ZeroMQ the right tool for this, or should I look for something like nanomsg?
Is PUB-SUB with forwarder the right pattern for this?
Or, am I missing something here?
Answers
Yes, ZeroMQ is capable of serving this need
Yes. ZeroMQ is a right tool ( rather a powerful tool-box of low-latency components ) for this. While nanomsg has a straight primitive for bus, the core distributed logic can be integrated in ZeroMQ framework
Yes & No. PUB-SUB as given above may serve for emulation of the "shout-cast"-to-bus and build on a SUB side-effect of using a subscription key(s). The WHOLE REST of the logic has to be re-thought and designed so as the whole scope of the fabrication meets your plans (ref. below). Also kindly bear in mind, that initial versions of ZeroMQ operated PUB/SUB primitive as "subscription filtering" of the incoming stream of messages being done on receiver side, so massive designs shall check against traffic-volumes / risk-of-flooding / process-inefficiency on the massive scale...
Yes. ZeroMQ is rather a well-tuned foundation of primitive elements ( as far as the architecture is discussed, not the power & performance thereof ) to build more clever, more robust & almost-linearly-scaleable Formal Communication Pattern(s). Do not get stuck to PUB/SUB or PAIR primitives once sketching Architecture. Any design will remain poor if one forgets where the True Powers comes from.
A good place to start a next step forward towards a scaleable & fault-resilient Bus
Thus a best next step one may do is IMHO to get a bit more global view, which may sound complicated for the first few things one tries to code with ZeroMQ, but if you at least jump to the page 265 of the Code Connected, Volume 1, if it were not the case of reading step-by-step thereto.
The fastest-ever learning-curve would be to have first an un-exposed view on the Fig.60 Republishing Updates and Fig.62 HA Clone Server pair for a possible High-availability approach and then go back to the roots, elements and details.
Here is what I ended up designing, if anyone is interested. Thanks everyone for the tips and pointers.
I have a message bus implemented using ZeroMQ (and CZMQ) running as a separate process.
The pattern is PUBLISHER-SUBSCRIBER with a LISTENER. They are connected using a PROXY.
In addition, there is a ROUTER invoked using a newly forked thread.
These three endpoints run on TCP and are bound to predefined ports which the clients know of.
PUBLISHER accepts all messages from clients.
SUBSCRIBER sends messages with a unique tag to the client who have subscribed to that tag.
LISTENER listens to all messages passing through. currently, this is for logging testing and purposes.
ROUTER provides a separate comms channel to clients. Messages such as control commands are directed here so that they will not get passed downstream.
Clients connect to,
PUBLISHER to send messages.
SUBSCRIBER to receive messages. Subscription is using unique tags.
ROUTER to send commands (check tag uniqueness etc.)
I am still doing implementation so there may be unseen problems, but right now it works fine. Also, there may be a more elegant way but I didn't want to throw away the PUB-SUB thing I had built.
Recently I've added some load-balancing capabilities to a piece of software that I wrote. It is a networked application that does some data crunching based on input coming from a SQL database. Since the crunching can be pretty intensive I've added the capability to have multiple instances of this application running on different servers to split the load but as it is now the load balancing is a manual act. A user must specify which instances take which portion of the input domain.
I would like to take that to the next level and program the instances to automatically negotiate the diving up of the input data and to recognize if one of them "disappears" (has crashed or has been powered down) so that the remaining instances can take on the failed instance's workload.
In order to implement this I'm considering using a simple heartbeat protocol between the instances to determine who's online and who isn't and while this is not terribly complicated I'd like to know if there are any established heartbeat network protocols (based on UDP, TCP or both).
Obviously this happens a lot in the networking world with clustering, fail-over and high-availability technologies so I guess in the end I'd like to know if maybe there are any established protocols or algorithms that I should be aware of or implement.
EDIT
It seems, based on the answers, that either there are no well established heart-beat protocols or that nobody knows about them (which would imply that they aren't so well established after all) in which case I'm just going to roll my own.
While none of the answers offered what I was looking for specifically I'm going to vote for Matt Davis's answer since it was the closest and he pointed out a good idea to use multicast.
Thank you all for your time~
Distribued Interactive Simulation (DIS), which is defined under IEEE Standard 1278, uses a default heartbeat of 5 seconds via UDP broadcast. A DIS heartbeat is essentially an Entity State PDU, which fully defines the state, including the position, of the given entity. Due to its application within the simulation community, DIS also uses a concept referred to as dead-reckoning to provide higher frequency heartbeats when the actual position, for example, is outside a given threshold of its predicted position.
In your case, a DIS Entity State PDU would be overkill. I only mention it to make note of the fact that heartbeats can vary in frequency depending on the circumstances. I don't know that you'd need something like this for the application you described, but you never know.
For heartbeats, use UDP, not TCP. A heartbeat is, by nature, a connectionless contrivance, so it goes that UDP (connectionless) is more relevant here than TCP (connection-oriented).
The thing to keep in mind about UDP broadcasts is that a broadcast message is confined to the broadcast domain. In short, if you have computers that are separated by a layer 3 device, e.g., a router, then broadcasts are not going to work because the router will not transmit broadcast messages from one broadcast domain to another. In this case, I would recommend using multicast since it will span the broadcast domains, providing the time-to-live (TTL) value is set high enough. It's also a more automated approach than directed unicast, which would require the sender to know the IP address of the receiver in order to send the message.
Broadcast a heartbeat every t using UDP; if you haven't heard from a machine in more than k*t, then it's assumed down. Be careful that the aggregate bandwidth used isn't a drain on resources. You can use IP broadcast addresses, or keep a list of specific IPs you're doing work for.
Make sure the heartbeat includes a "reboot count" as well as "machine ID" so that you know previous server state isn't around.
I'd recommend using MapReduce if it fits. It would save a lot of work.
I'm not sure this will answer the question but you might be interested by the way Weblogic Server clustering work under the hood. From the book Mastering BEA WebLogic Server:
[...] WebLogic Server clustering provides a loose coupling of the servers in the cluster. Each server in the cluster is independent and does not rely on any other server for any fundamental operations. Even if contact with every other server is lost, each server will continue to run and be able to process the requests it receives. Each server in the cluster maintains its own list of other servers in the cluster through periodic heartbeat messages. Every 10 seconds, each server sends a heartbeat message to the other servers in the cluster to let them know it is still alive. Heartbeat messages are sent using IP multicast technology built into the JVM, making this mechanism efficient and scalable as the number of servers in the cluster gets large. Each server receives these heartbeat messages from other servers and uses them to maintain its current cluster membership list. If a server misses receiving three heartbeat messages in a row from any other server, it takes that server out of its membership list until it receives another heartbeat message from that server. This heartbeat technology allows servers to be dynamically added and dropped from the cluster with no impact on the existing servers’ configurations.
Cisco content switches are a hardware solution for this problem. They implement a virtual IP address as a front end to multiple real servers, whose real IP addresses are known to the switch. The switch periodically sends HTTP HEAD requests to the web servers, to verify they are still running (which the switch software calls a "keepalive", although this doesn't keep the server itself alive). The Cisco switch accepts traffic on the virtual IP and forwards it to the actual web servers, using configurable load balancing such as round-robin, or user-defined load balancing.
These switches retail in the $3-10K range, although my business partner picked one up on eBay for about $300 a year ago. If you can afford one, they do represent a proven hardware solution to the question of how to have a service spread transparently across multiple servers. Redhat includes a built-in port configuration so that you could implement your own Cisco switch using a cheap RedHat box. Google for "virtual ip address" and "cisco content router" for more information.
In addition to trying hardware load-balancers, you can also try a free-open-source load-balancing software application such as HAProxy, available for Linux and the BSDs.