samza-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Yi Pan <nickpa...@gmail.com>
Subject Re: Thoughts and obesrvations on Samza
Date Thu, 02 Jul 2015 22:38:10 GMT
@Guozhang, yes, that's what I meant. From Kafka consumers' point of view,
it pretty much boils down to answer the following question:
1. For Kafka consumer in each container (i.e. a Samza worker), which topic
partitions it should consume.
Samza's current StreamTask model still makes sense to me and the
partition-to-task mapping is the debatable point that whether that should
be in Kafka or stays in a separate module. As we discussed earlier, some
simple partition-to-task mapping maybe expressed as co-partition
distribution among different topics in Kafka (forgive me if I had make
mistakes here since I am not 100% sure about how partition distribution
policies work in Kafka). However, more complex and application-specific
partition-to-task mapping would need to stay outside of Kafka. One example
is the discussion on SQL tasks:
https://issues.apache.org/jira/browse/SAMZA-516?focusedCommentId=14288685&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-14288685

On Thu, Jul 2, 2015 at 2:47 PM, Guozhang Wang <wangguoz@gmail.com> wrote:

> Since the resource scheduling systems like YARN / Mesos only gives Samza a
> couple of resource units (or "containers") to run processes, while Samza
> still needs to handle task assignment / scheduling like which tasks should
> be allocated to which containers that consume from which partitions, etc. I
> think this is want Yi meant for "partition management"?
>
> On Thu, Jul 2, 2015 at 2:35 PM, Yi Pan <nickpan47@gmail.com> wrote:
>
> > @Jay, yes, the current function in the JobCoordinator is just partition
> > management. Maybe we should just call it PartitionManager to make it
> > explicit.
> >
> > -Yi
> >
> > On Thu, Jul 2, 2015 at 2:24 PM, Jay Kreps <jay@confluent.io> wrote:
> >
> > > Hey Yi,
> > >
> > > What does the JobCoordinator do? YARN/Mesos/etc would be doing the
> actual
> > > resource assignment, process restart, etc, right? Is the additional
> value
> > > add of the JobCoordinator just partition management?
> > >
> > > -Jay
> > >
> > > On Thu, Jul 2, 2015 at 11:32 AM, Yi Pan <nickpan47@gmail.com> wrote:
> > >
> > > > Hi, all,
> > > >
> > > >
> > > > Thanks Chris for sending out this proposal and Jay for sharing the
> > > > extremely illustrative prototype code.
> > > >
> > > >
> > > > I have been thinking it over many times and want to list out my
> > personal
> > > > opinions below:
> > > >
> > > > 1. Generally, I agree with most of the people here on the mailing
> list
> > on
> > > > two points:
> > > >
> > > >    a. Deeper integration w/ Kafka is great. No more confusing mapping
> > > from
> > > > SystemStreamPartition to TopicPartition etc.
> > > >
> > > >    b. Separation the ingestion vs transformation greatly simplify the
> > > > systems APIs
> > > >
> > > > Having the above two changes would allow us to remove many
> unnecessary
> > > > complexities introduced by those pluggable interfaces Chris’ pointed
> > out,
> > > > e.g. pluggable streaming systems and serde.
> > > >
> > > >
> > > > To recall one of Chris’s statement on difficulties in dynamic
> > > deployment, I
> > > > believe that the difficulties are mainly the result of tight-coupling
> > of
> > > > partition assignment vs the container deployment in the current
> system.
> > > The
> > > > current container deployment requires a pre-defined partition
> > assignment
> > > > strategy coupled together w/ the deployment configuration before we
> can
> > > > submit to YARN and start the Samza container, which makes the
> launching
> > > > process super long. Also, fault-tolerance and the embedded
> > JobCoordinator
> > > > code in YARN AppMaster is another way of  making dynamic deployment
> > more
> > > > complex and difficult.
> > > >
> > > >
> > > > First, borrowing Yan’s term, let’s call the Samza standalone process
> a
> > > > Samza worker. Here is what I have been thinking:
> > > >
> > > > 1. Separate the execution framework from partition assignment/load
> > > > balancing:
> > > >
> > > >     a. a Samza worker should be launched by execution framework that
> > only
> > > > deals w/ process placement to available nodes. The execution
> framework
> > > now
> > > > should only deal w/ how many such processes are needed, where to put
> > > them,
> > > > and how to keep them alive.
> > > >
> > > >     b. Partition assignment/load balancing can be a pluggable
> interface
> > > in
> > > > Samza that allows the Samza workers to ask for partition assignments.
> > > Let’s
> > > > borrow the name JobCoordinator for now. To allow fault-tolerance in
> > case
> > > of
> > > > failure, the partition assignments to workers need to be dynamic.
> > Hence,
> > > > the abstract interface would be much like what Jay’s code illustrate:
> > > > get()/onAssigned()/onRevoke(). The implementation of the partition
> > > > assignment can be either:
> > > >
> > > >         a) completely rely on Kafka.
> > > >
> > > >         b) explicit partition assignment via JobCoordinator. Chris’s
> > work
> > > > in SAMZA-516 can be easily incorporated here. The use case in
> SAMZA-41
> > > that
> > > > runs Samza ProcessJob w/ static partition assignment can be
> implemented
> > > of
> > > > JobCoordinator via any home-grown implementation of distributed
> > > > coordinator. All the work we did in LinkedIn to support dynamic
> > partition
> > > > assignment and host-affinity SAMZA-617 can be nicely reused as an
> > > > implementation of JobCoordinator.
> > > >
> > > >
> > > > When we did the above work, I can see three usage patterns in Samza:
> > > >
> > > >    a. Samza as a library: Samza has a set of libraries to provide
> > stream
> > > > processing, just like a third Kafka client (as illustrated in Jay’s
> > > > example). The execution/deployment is totally controlled by the
> > > application
> > > > and the partition coordination is done via Kafka
> > > >
> > > >    b. Samza as a process: Samza runs as a standalone process. There
> may
> > > not
> > > > be a execution framework to launch and deploy Samza processes. The
> > > > partition assignment is pluggable JobCoordinator.
> > > >
> > > >    c. Samza as a service: Samza runs as a collection of processes.
> > There
> > > > will be an execution framework to allocate resource, launch and
> deploy
> > > > Samza workers and keep them alive. The same pluggable JobCoordinator
> is
> > > > desirable here as well.
> > > >
> > > >
> > > > Lastly, I would argue that CopyCat in KIP-26 should probably follow
> the
> > > > same model. Hence, in Samza as a service model as in LinkedIn, we can
> > use
> > > > the same fault tolerance execution framework to run CopyCat and Samza
> > w/o
> > > > the need to operate two service platforms, which should address
> > Sriram’s
> > > > comment in the email thread.
> > > >
> > > >
> > > > Hope the above makes sense. Thanks all!
> > > >
> > > >
> > > > -Yi
> > > >
> > > > On Thu, Jul 2, 2015 at 9:53 AM, Sriram <sriram.sub@gmail.com> wrote:
> > > >
> > > > > One thing that is worth exploring is to have a transformation and
> > > > > ingestion library in Kafka but use the same framework for fault
> > > > tolerance,
> > > > > resource isolation and management. The biggest difference I see in
> > > these
> > > > > two use cases is the API and data model.
> > > > >
> > > > >
> > > > > > On Jul 2, 2015, at 8:59 AM, Jay Kreps <jay@confluent.io> wrote:
> > > > > >
> > > > > > Hey Garry,
> > > > > >
> > > > > > Yeah that's super frustrating. I'd be happy to chat more about
> this
> > > if
> > > > > > you'd be interested. I think Chris and I started with the idea of
> > > "what
> > > > > > would it take to make Samza a kick-ass ingestion tool" but
> > ultimately
> > > > we
> > > > > > kind of came around to the idea that ingestion and transformation
> > had
> > > > > > pretty different needs and coupling the two made things hard.
> > > > > >
> > > > > > For what it's worth I think copycat (KIP-26) actually will do
> what
> > > you
> > > > > are
> > > > > > looking for.
> > > > > >
> > > > > > With regard to your point about slider, I don't necessarily
> > disagree.
> > > > > But I
> > > > > > think getting good YARN support is quite doable and I think we
> can
> > > make
> > > > > > that work well. I think the issue this proposal solves is that
> > > > > technically
> > > > > > it is pretty hard to support multiple cluster management systems
> > the
> > > > way
> > > > > > things are now, you need to write an "app master" or "framework"
> > for
> > > > each
> > > > > > and they are all a little different so testing is really hard. In
> > the
> > > > > > absence of this we have been stuck with just YARN which has
> > fantastic
> > > > > > penetration in the Hadoopy part of the org, but zero penetration
> > > > > elsewhere.
> > > > > > Given the huge amount of work being put in to slider, marathon,
> aws
> > > > > > tooling, not to mention the umpteen related packaging
> technologies
> > > > people
> > > > > > want to use (Docker, Kubernetes, various cloud-specific deploy
> > tools,
> > > > > etc)
> > > > > > I really think it is important to get this right.
> > > > > >
> > > > > > -Jay
> > > > > >
> > > > > > On Thu, Jul 2, 2015 at 4:17 AM, Garry Turkington <
> > > > > > g.turkington@improvedigital.com> wrote:
> > > > > >
> > > > > >> Hi all,
> > > > > >>
> > > > > >> I think the question below re does Samza become a sub-project of
> > > Kafka
> > > > > >> highlights the broader point around migration. Chris mentions
> > > Samza's
> > > > > >> maturity is heading towards a v1 release but I'm not sure it
> feels
> > > > > right to
> > > > > >> launch a v1 then immediately plan to deprecate most of it.
> > > > > >>
> > > > > >> From a selfish perspective I have some guys who have started
> > working
> > > > > with
> > > > > >> Samza and building some new consumers/producers was next up.
> > Sounds
> > > > like
> > > > > >> that is absolutely not the direction to go. I need to look into
> > the
> > > > KIP
> > > > > in
> > > > > >> more detail but for me the attractiveness of adding new Samza
> > > > > >> consumer/producers -- even if yes all they were doing was really
> > > > getting
> > > > > >> data into and out of Kafka --  was to avoid  having to worry
> about
> > > the
> > > > > >> lifecycle management of external clients. If there is a generic
> > > Kafka
> > > > > >> ingress/egress layer that I can plug a new connector into and
> > have a
> > > > > lot of
> > > > > >> the heavy lifting re scale and reliability done for me then it
> > gives
> > > > me
> > > > > all
> > > > > >> the pushing new consumers/producers would. If not then it
> > > complicates
> > > > my
> > > > > >> operational deployments.
> > > > > >>
> > > > > >> Which is similar to my other question with the proposal -- if we
> > > > build a
> > > > > >> fully available/stand-alone Samza plus the requisite shims to
> > > > integrate
> > > > > >> with Slider etc I suspect the former may be a lot more work than
> > we
> > > > > think.
> > > > > >> We may make it much easier for a newcomer to get something
> running
> > > but
> > > > > >> having them step up and get a reliable production deployment may
> > > still
> > > > > >> dominate mailing list  traffic, if for different reasons than
> > today.
> > > > > >>
> > > > > >> Don't get me wrong -- I'm comfortable with making the Samza
> > > dependency
> > > > > on
> > > > > >> Kafka much more explicit and I absolutely see the benefits  in
> the
> > > > > >> reduction of duplication and clashing terminologies/abstractions
> > > that
> > > > > >> Chris/Jay describe. Samza as a library would likely be a very
> nice
> > > > tool
> > > > > to
> > > > > >> add to the Kafka ecosystem. I just have the concerns above re
> the
> > > > > >> operational side.
> > > > > >>
> > > > > >> Garry
> > > > > >>
> > > > > >> -----Original Message-----
> > > > > >> From: Gianmarco De Francisci Morales [mailto:gdfm@apache.org]
> > > > > >> Sent: 02 July 2015 12:56
> > > > > >> To: dev@samza.apache.org
> > > > > >> Subject: Re: Thoughts and obesrvations on Samza
> > > > > >>
> > > > > >> Very interesting thoughts.
> > > > > >> From outside, I have always perceived Samza as a computing layer
> > > over
> > > > > >> Kafka.
> > > > > >>
> > > > > >> The question, maybe a bit provocative, is "should Samza be a
> > > > sub-project
> > > > > >> of Kafka then?"
> > > > > >> Or does it make sense to keep it as a separate project with a
> > > separate
> > > > > >> governance?
> > > > > >>
> > > > > >> Cheers,
> > > > > >>
> > > > > >> --
> > > > > >> Gianmarco
> > > > > >>
> > > > > >>> On 2 July 2015 at 08:59, Yan Fang <yanfang724@gmail.com>
> wrote:
> > > > > >>>
> > > > > >>> Overall, I agree to couple with Kafka more tightly. Because
> Samza
> > > de
> > > > > >>> facto is based on Kafka, and it should leverage what Kafka has.
> > At
> > > > the
> > > > > >>> same time, Kafka does not need to reinvent what Samza already
> > has.
> > > I
> > > > > >>> also like the idea of separating the ingestion and
> > transformation.
> > > > > >>>
> > > > > >>> But it is a little difficult for me to image how the Samza will
> > > look
> > > > > >> like.
> > > > > >>> And I feel Chris and Jay have a little difference in terms of
> how
> > > > > >>> Samza should look like.
> > > > > >>>
> > > > > >>> *** Will it look like what Jay's code shows (A client of
> Kakfa) ?
> > > And
> > > > > >>> user's application code calls this client?
> > > > > >>>
> > > > > >>> 1. If we make Samza be a library of Kafka (like what the code
> > > shows),
> > > > > >>> how do we implement auto-balance and fault-tolerance? Are they
> > > taken
> > > > > >>> care by the Kafka broker or other mechanism, such as "Samza
> > worker"
> > > > > >>> (just make up the name) ?
> > > > > >>>
> > > > > >>> 2. What about other features, such as auto-scaling, shared
> state,
> > > > > >>> monitoring?
> > > > > >>>
> > > > > >>>
> > > > > >>> *** If we have Samza standalone, (is this what Chris suggests?)
> > > > > >>>
> > > > > >>> 1. we still need to ingest data from Kakfa and produce to it.
> > Then
> > > it
> > > > > >>> becomes the same as what Samza looks like now, except it does
> not
> > > > rely
> > > > > >>> on Yarn anymore.
> > > > > >>>
> > > > > >>> 2. if it is standalone, how can it leverage Kafka's metrics,
> > logs,
> > > > > >>> etc? Use Kafka code as the dependency?
> > > > > >>>
> > > > > >>>
> > > > > >>> Thanks,
> > > > > >>>
> > > > > >>> Fang, Yan
> > > > > >>> yanfang724@gmail.com
> > > > > >>>
> > > > > >>>> On Wed, Jul 1, 2015 at 5:46 PM, Guozhang Wang <
> > wangguoz@gmail.com
> > > >
> > > > > >>> wrote:
> > > > > >>>
> > > > > >>>> Read through the code example and it looks good to me. A few
> > > > > >>>> thoughts regarding deployment:
> > > > > >>>>
> > > > > >>>> Today Samza deploys as executable runnable like:
> > > > > >>>>
> > > > > >>>> deploy/samza/bin/run-job.sh --config-factory=...
> > > > > >> --config-path=file://...
> > > > > >>>>
> > > > > >>>> And this proposal advocate for deploying Samza more as
> embedded
> > > > > >>>> libraries in user application code (ignoring the terminology
> > since
> > > > > >>>> it is not the
> > > > > >>> same
> > > > > >>>> as the prototype code):
> > > > > >>>>
> > > > > >>>> StreamTask task = new MyStreamTask(configs); Thread thread =
> new
> > > > > >>>> Thread(task); thread.start();
> > > > > >>>>
> > > > > >>>> I think both of these deployment modes are important for
> > different
> > > > > >>>> types
> > > > > >>> of
> > > > > >>>> users. That said, I think making Samza purely standalone is
> > still
> > > > > >>>> sufficient for either runnable or library modes.
> > > > > >>>>
> > > > > >>>> Guozhang
> > > > > >>>>
> > > > > >>>>> On Tue, Jun 30, 2015 at 11:33 PM, Jay Kreps <
> jay@confluent.io>
> > > > > wrote:
> > > > > >>>>>
> > > > > >>>>> Looks like gmail mangled the code example, it was supposed to
> > > look
> > > > > >>>>> like
> > > > > >>>>> this:
> > > > > >>>>>
> > > > > >>>>> Properties props = new Properties();
> > > > > >>>>> props.put("bootstrap.servers", "localhost:4242");
> > StreamingConfig
> > > > > >>>>> config = new StreamingConfig(props);
> > > > > >>>>> config.subscribe("test-topic-1", "test-topic-2");
> > > > > >>>>> config.processor(ExampleStreamProcessor.class);
> > > > > >>>>> config.serialization(new StringSerializer(), new
> > > > > >>>>> StringDeserializer()); KafkaStreaming container = new
> > > > > >>>>> KafkaStreaming(config); container.run();
> > > > > >>>>>
> > > > > >>>>> -Jay
> > > > > >>>>>
> > > > > >>>>> On Tue, Jun 30, 2015 at 11:32 PM, Jay Kreps <
> jay@confluent.io>
> > > > > >> wrote:
> > > > > >>>>>
> > > > > >>>>>> Hey guys,
> > > > > >>>>>>
> > > > > >>>>>> This came out of some conversations Chris and I were having
> > > > > >>>>>> around
> > > > > >>>>> whether
> > > > > >>>>>> it would make sense to use Samza as a kind of data ingestion
> > > > > >>> framework
> > > > > >>>>> for
> > > > > >>>>>> Kafka (which ultimately lead to KIP-26 "copycat"). This kind
> > of
> > > > > >>>> combined
> > > > > >>>>>> with complaints around config and YARN and the discussion
> > around
> > > > > >>>>>> how
> > > > > >>> to
> > > > > >>>>>> best do a standalone mode.
> > > > > >>>>>>
> > > > > >>>>>> So the thought experiment was, given that Samza was
> basically
> > > > > >>>>>> already totally Kafka specific, what if you just embraced
> that
> > > > > >>>>>> and turned it
> > > > > >>>> into
> > > > > >>>>>> something less like a heavyweight framework and more like a
> > > > > >>>>>> third
> > > > > >>> Kafka
> > > > > >>>>>> client--a kind of "producing consumer" with state management
> > > > > >>>> facilities.
> > > > > >>>>>> Basically a library. Instead of a complex stream processing
> > > > > >>>>>> framework
> > > > > >>>>> this
> > > > > >>>>>> would actually be a very simple thing, not much more
> > complicated
> > > > > >>>>>> to
> > > > > >>> use
> > > > > >>>>> or
> > > > > >>>>>> operate than a Kafka consumer. As Chris said we thought
> about
> > it
> > > > > >>>>>> a
> > > > > >>> lot
> > > > > >>>> of
> > > > > >>>>>> what Samza (and the other stream processing systems were
> > doing)
> > > > > >>> seemed
> > > > > >>>>> like
> > > > > >>>>>> kind of a hangover from MapReduce.
> > > > > >>>>>>
> > > > > >>>>>> Of course you need to ingest/output data to and from the
> > stream
> > > > > >>>>>> processing. But when we actually looked into how that would
> > > > > >>>>>> work,
> > > > > >>> Samza
> > > > > >>>>>> isn't really an ideal data ingestion framework for a bunch
> of
> > > > > >>> reasons.
> > > > > >>>> To
> > > > > >>>>>> really do that right you need a pretty different internal
> data
> > > > > >>>>>> model
> > > > > >>>> and
> > > > > >>>>>> set of apis. So what if you split them and had an api for
> > Kafka
> > > > > >>>>>> ingress/egress (copycat AKA KIP-26) and a separate api for
> > Kafka
> > > > > >>>>>> transformation (Samza).
> > > > > >>>>>>
> > > > > >>>>>> This would also allow really embracing the same terminology
> > and
> > > > > >>>>>> conventions. One complaint about the current state is that
> the
> > > > > >>>>>> two
> > > > > >>>>> systems
> > > > > >>>>>> kind of feel bolted on. Terminology like "stream" vs "topic"
> > and
> > > > > >>>>> different
> > > > > >>>>>> config and monitoring systems means you kind of have to
> learn
> > > > > >>>>>> Kafka's
> > > > > >>>>> way,
> > > > > >>>>>> then learn Samza's slightly different way, then kind of
> > > > > >>>>>> understand
> > > > > >>> how
> > > > > >>>>> they
> > > > > >>>>>> map to each other, which having walked a few people through
> > this
> > > > > >>>>>> is surprisingly tricky for folks to get.
> > > > > >>>>>>
> > > > > >>>>>> Since I have been spending a lot of time on airplanes I
> hacked
> > > > > >>>>>> up an ernest but still somewhat incomplete prototype of what
> > > > > >>>>>> this would
> > > > > >>> look
> > > > > >>>>>> like. This is just unceremoniously dumped into Kafka as it
> > > > > >>>>>> required a
> > > > > >>>> few
> > > > > >>>>>> changes to the new consumer. Here is the code:
> > > > > >>>
> > > >
> https://github.com/jkreps/kafka/tree/streams/clients/src/main/java/org
> > > > > >>> /apache/kafka/clients/streaming
> > > > > >>>>>>
> > > > > >>>>>> For the purpose of the prototype I just liberally renamed
> > > > > >>>>>> everything
> > > > > >>> to
> > > > > >>>>>> try to align it with Kafka with no regard for compatibility.
> > > > > >>>>>>
> > > > > >>>>>> To use this would be something like this:
> > > > > >>>>>> Properties props = new Properties();
> > > > > >>>>>> props.put("bootstrap.servers", "localhost:4242");
> > > > > >>>>>> StreamingConfig config = new
> > > > > >>> StreamingConfig(props);
> > > > > >>>>> config.subscribe("test-topic-1",
> > > > > >>>>>> "test-topic-2");
> > config.processor(ExampleStreamProcessor.class);
> > > > > >>>>> config.serialization(new
> > > > > >>>>>> StringSerializer(), new StringDeserializer());
> KafkaStreaming
> > > > > >>>> container =
> > > > > >>>>>> new KafkaStreaming(config); container.run();
> > > > > >>>>>>
> > > > > >>>>>> KafkaStreaming is basically the SamzaContainer;
> > StreamProcessor
> > > > > >>>>>> is basically StreamTask.
> > > > > >>>>>>
> > > > > >>>>>> So rather than putting all the class names in a file and
> then
> > > > > >>>>>> having
> > > > > >>>> the
> > > > > >>>>>> job assembled by reflection, you just instantiate the
> > container
> > > > > >>>>>> programmatically. Work is balanced over however many
> instances
> > > > > >>>>>> of
> > > > > >>> this
> > > > > >>>>> are
> > > > > >>>>>> alive at any time (i.e. if an instance dies, new tasks are
> > added
> > > > > >>>>>> to
> > > > > >>> the
> > > > > >>>>>> existing containers without shutting them down).
> > > > > >>>>>>
> > > > > >>>>>> We would provide some glue for running this stuff in YARN
> via
> > > > > >>>>>> Slider, Mesos via Marathon, and AWS using some of their
> tools
> > > > > >>>>>> but from the
> > > > > >>>> point
> > > > > >>>>> of
> > > > > >>>>>> view of these frameworks these stream processing jobs are
> just
> > > > > >>>> stateless
> > > > > >>>>>> services that can come and go and expand and contract at
> will.
> > > > > >>>>>> There
> > > > > >>> is
> > > > > >>>>> no
> > > > > >>>>>> more custom scheduler.
> > > > > >>>>>>
> > > > > >>>>>> Here are some relevant details:
> > > > > >>>>>>
> > > > > >>>>>>   1. It is only ~1300 lines of code, it would get larger if
> we
> > > > > >>>>>>   productionized but not vastly larger. We really do get a
> ton
> > > > > >>>>>> of
> > > > > >>>>> leverage
> > > > > >>>>>>   out of Kafka.
> > > > > >>>>>>   2. Partition management is fully delegated to the new
> > > consumer.
> > > > > >>> This
> > > > > >>>>>>   is nice since now any partition management strategy
> > available
> > > > > >>>>>> to
> > > > > >>>> Kafka
> > > > > >>>>>>   consumer is also available to Samza (and vice versa) and
> > with
> > > > > >>>>>> the
> > > > > >>>>> exact
> > > > > >>>>>>   same configs.
> > > > > >>>>>>   3. It supports state as well as state reuse
> > > > > >>>>>>
> > > > > >>>>>> Anyhow take a look, hopefully it is thought provoking.
> > > > > >>>>>>
> > > > > >>>>>> -Jay
> > > > > >>>>>>
> > > > > >>>>>>
> > > > > >>>>>>
> > > > > >>>>>> On Tue, Jun 30, 2015 at 6:55 PM, Chris Riccomini <
> > > > > >>>> criccomini@apache.org>
> > > > > >>>>>> wrote:
> > > > > >>>>>>
> > > > > >>>>>>> Hey all,
> > > > > >>>>>>>
> > > > > >>>>>>> I have had some discussions with Samza engineers at
> LinkedIn
> > > > > >>>>>>> and
> > > > > >>>>> Confluent
> > > > > >>>>>>> and we came up with a few observations and would like to
> > > > > >>>>>>> propose
> > > > > >>> some
> > > > > >>>>>>> changes.
> > > > > >>>>>>>
> > > > > >>>>>>> We've observed some things that I want to call out about
> > > > > >>>>>>> Samza's
> > > > > >>>> design,
> > > > > >>>>>>> and I'd like to propose some changes.
> > > > > >>>>>>>
> > > > > >>>>>>> * Samza is dependent upon a dynamic deployment system.
> > > > > >>>>>>> * Samza is too pluggable.
> > > > > >>>>>>> * Samza's SystemConsumer/SystemProducer and Kafka's
> consumer
> > > > > >>>>>>> APIs
> > > > > >>> are
> > > > > >>>>>>> trying to solve a lot of the same problems.
> > > > > >>>>>>>
> > > > > >>>>>>> All three of these issues are related, but I'll address
> them
> > in
> > > > > >>> order.
> > > > > >>>>>>>
> > > > > >>>>>>> Deployment
> > > > > >>>>>>>
> > > > > >>>>>>> Samza strongly depends on the use of a dynamic deployment
> > > > > >>>>>>> scheduler
> > > > > >>>> such
> > > > > >>>>>>> as
> > > > > >>>>>>> YARN, Mesos, etc. When we initially built Samza, we bet
> that
> > > > > >>>>>>> there
> > > > > >>>> would
> > > > > >>>>>>> be
> > > > > >>>>>>> one or two winners in this area, and we could support them,
> > and
> > > > > >>>>>>> the
> > > > > >>>> rest
> > > > > >>>>>>> would go away. In reality, there are many variations.
> > > > > >>>>>>> Furthermore,
> > > > > >>>> many
> > > > > >>>>>>> people still prefer to just start their processors like
> > normal
> > > > > >>>>>>> Java processes, and use traditional deployment scripts such
> > as
> > > > > >>>>>>> Fabric,
> > > > > >>>> Chef,
> > > > > >>>>>>> Ansible, etc. Forcing a deployment system on users makes
> the
> > > > > >>>>>>> Samza start-up process really painful for first time users.
> > > > > >>>>>>>
> > > > > >>>>>>> Dynamic deployment as a requirement was also a bit of a
> > > > > >>>>>>> mis-fire
> > > > > >>>> because
> > > > > >>>>>>> of
> > > > > >>>>>>> a fundamental misunderstanding between the nature of batch
> > jobs
> > > > > >>>>>>> and
> > > > > >>>>> stream
> > > > > >>>>>>> processing jobs. Early on, we made conscious effort to
> favor
> > > > > >>>>>>> the
> > > > > >>>> Hadoop
> > > > > >>>>>>> (Map/Reduce) way of doing things, since it worked and was
> > well
> > > > > >>>>> understood.
> > > > > >>>>>>> One thing that we missed was that batch jobs have a
> definite
> > > > > >>>> beginning,
> > > > > >>>>>>> and
> > > > > >>>>>>> end, and stream processing jobs don't (usually). This leads
> > to
> > > > > >>>>>>> a
> > > > > >>> much
> > > > > >>>>>>> simpler scheduling problem for stream processors. You
> > basically
> > > > > >>>>>>> just
> > > > > >>>>> need
> > > > > >>>>>>> to find a place to start the processor, and start it. The
> way
> > > > > >>>>>>> we run grids, at LinkedIn, there's no concept of a cluster
> > > > > >>>>>>> being "full". We always
> > > > > >>>> add
> > > > > >>>>>>> more machines. The problem with coupling Samza with a
> > scheduler
> > > > > >>>>>>> is
> > > > > >>>> that
> > > > > >>>>>>> Samza (as a framework) now has to handle deployment. This
> > pulls
> > > > > >>>>>>> in a
> > > > > >>>>> bunch
> > > > > >>>>>>> of things such as configuration distribution (config
> stream),
> > > > > >>>>>>> shell
> > > > > >>>>> scrips
> > > > > >>>>>>> (bin/run-job.sh, JobRunner), packaging (all the .tgz
> stuff),
> > > etc.
> > > > > >>>>>>>
> > > > > >>>>>>> Another reason for requiring dynamic deployment was to
> > support
> > > > > >>>>>>> data locality. If you want to have locality, you need to
> put
> > > > > >>>>>>> your
> > > > > >>>> processors
> > > > > >>>>>>> close to the data they're processing. Upon further
> > > > > >>>>>>> investigation,
> > > > > >>>>> though,
> > > > > >>>>>>> this feature is not that beneficial. There is some good
> > > > > >>>>>>> discussion
> > > > > >>>> about
> > > > > >>>>>>> some problems with it on SAMZA-335. Again, we took the
> > > > > >>>>>>> Map/Reduce
> > > > > >>>> path,
> > > > > >>>>>>> but
> > > > > >>>>>>> there are some fundamental differences between HDFS and
> > Kafka.
> > > > > >>>>>>> HDFS
> > > > > >>>> has
> > > > > >>>>>>> blocks, while Kafka has partitions. This leads to less
> > > > > >>>>>>> optimization potential with stream processors on top of
> > Kafka.
> > > > > >>>>>>>
> > > > > >>>>>>> This feature is also used as a crutch. Samza doesn't have
> any
> > > > > >>>>>>> built
> > > > > >>> in
> > > > > >>>>>>> fault-tolerance logic. Instead, it depends on the dynamic
> > > > > >>>>>>> deployment scheduling system to handle restarts when a
> > > > > >>>>>>> processor dies. This has
> > > > > >>>>> made
> > > > > >>>>>>> it very difficult to write a standalone Samza container
> > > > > >> (SAMZA-516).
> > > > > >>>>>>>
> > > > > >>>>>>> Pluggability
> > > > > >>>>>>>
> > > > > >>>>>>> In some cases pluggability is good, but I think that we've
> > gone
> > > > > >>>>>>> too
> > > > > >>>> far
> > > > > >>>>>>> with it. Currently, Samza has:
> > > > > >>>>>>>
> > > > > >>>>>>> * Pluggable config.
> > > > > >>>>>>> * Pluggable metrics.
> > > > > >>>>>>> * Pluggable deployment systems.
> > > > > >>>>>>> * Pluggable streaming systems (SystemConsumer,
> > SystemProducer,
> > > > > >> etc).
> > > > > >>>>>>> * Pluggable serdes.
> > > > > >>>>>>> * Pluggable storage engines.
> > > > > >>>>>>> * Pluggable strategies for just about every component
> > > > > >>> (MessageChooser,
> > > > > >>>>>>> SystemStreamPartitionGrouper, ConfigRewriter, etc).
> > > > > >>>>>>>
> > > > > >>>>>>> There's probably more that I've forgotten, as well. Some of
> > > > > >>>>>>> these
> > > > > >>> are
> > > > > >>>>>>> useful, but some have proven not to be. This all comes at a
> > > cost:
> > > > > >>>>>>> complexity. This complexity is making it harder for our
> users
> > > > > >>>>>>> to
> > > > > >>> pick
> > > > > >>>> up
> > > > > >>>>>>> and use Samza out of the box. It also makes it difficult
> for
> > > > > >>>>>>> Samza developers to reason about what the characteristics
> of
> > > > > >>>>>>> the container (since the characteristics change depending
> on
> > > > > >>>>>>> which plugins are use).
> > > > > >>>>>>>
> > > > > >>>>>>> The issues with pluggability are most visible in the System
> > > APIs.
> > > > > >>> What
> > > > > >>>>>>> Samza really requires to be functional is Kafka as its
> > > > > >>>>>>> transport
> > > > > >>>> layer.
> > > > > >>>>>>> But
> > > > > >>>>>>> we've conflated two unrelated use cases into one API:
> > > > > >>>>>>>
> > > > > >>>>>>> 1. Get data into/out of Kafka.
> > > > > >>>>>>> 2. Process the data in Kafka.
> > > > > >>>>>>>
> > > > > >>>>>>> The current System API supports both of these use cases.
> The
> > > > > >>>>>>> problem
> > > > > >>>> is,
> > > > > >>>>>>> we
> > > > > >>>>>>> actually want different features for each use case. By
> > papering
> > > > > >>>>>>> over
> > > > > >>>>> these
> > > > > >>>>>>> two use cases, and providing a single API, we've
> introduced a
> > > > > >>>>>>> ton of
> > > > > >>>>> leaky
> > > > > >>>>>>> abstractions.
> > > > > >>>>>>>
> > > > > >>>>>>> For example, what we'd really like in (2) is to have
> > > > > >>>>>>> monotonically increasing longs for offsets (like Kafka).
> This
> > > > > >>>>>>> would be at odds
> > > > > >>> with
> > > > > >>>>> (1),
> > > > > >>>>>>> though, since different systems have different
> > > > > >>>>> SCNs/Offsets/UUIDs/vectors.
> > > > > >>>>>>> There was discussion both on the mailing list and the SQL
> > JIRAs
> > > > > >>> about
> > > > > >>>>> the
> > > > > >>>>>>> need for this.
> > > > > >>>>>>>
> > > > > >>>>>>> The same thing holds true for replayability. Kafka allows
> us
> > to
> > > > > >>> rewind
> > > > > >>>>>>> when
> > > > > >>>>>>> we have a failure. Many other systems don't. In some cases,
> > > > > >>>>>>> systems
> > > > > >>>>> return
> > > > > >>>>>>> null for their offsets (e.g. WikipediaSystemConsumer)
> because
> > > > > >>>>>>> they
> > > > > >>>> have
> > > > > >>>>> no
> > > > > >>>>>>> offsets.
> > > > > >>>>>>>
> > > > > >>>>>>> Partitioning is another example. Kafka supports
> partitioning,
> > > > > >>>>>>> but
> > > > > >>> many
> > > > > >>>>>>> systems don't. We model this by having a single partition
> for
> > > > > >>>>>>> those systems. Still, other systems model partitioning
> > > > > >> differently (e.g.
> > > > > >>>>>>> Kinesis).
> > > > > >>>>>>>
> > > > > >>>>>>> The SystemAdmin interface is also a mess. Creating streams
> > in a
> > > > > >>>>>>> system-agnostic way is almost impossible. As is modeling
> > > > > >>>>>>> metadata
> > > > > >>> for
> > > > > >>>>> the
> > > > > >>>>>>> system (replication factor, partitions, location, etc). The
> > > > > >>>>>>> list
> > > > > >>> goes
> > > > > >>>>> on.
> > > > > >>>>>>>
> > > > > >>>>>>> Duplicate work
> > > > > >>>>>>>
> > > > > >>>>>>> At the time that we began writing Samza, Kafka's consumer
> and
> > > > > >>> producer
> > > > > >>>>>>> APIs
> > > > > >>>>>>> had a relatively weak feature set. On the consumer-side,
> you
> > > > > >>>>>>> had two
> > > > > >>>>>>> options: use the high level consumer, or the simple
> consumer.
> > > > > >>>>>>> The
> > > > > >>>>> problem
> > > > > >>>>>>> with the high-level consumer was that it controlled your
> > > > > >>>>>>> offsets, partition assignments, and the order in which you
> > > > > >>>>>>> received messages. The
> > > > > >>> problem
> > > > > >>>>>>> with
> > > > > >>>>>>> the simple consumer is that it's not simple. It's basic.
> You
> > > > > >>>>>>> end up
> > > > > >>>>> having
> > > > > >>>>>>> to handle a lot of really low-level stuff that you
> shouldn't.
> > > > > >>>>>>> We
> > > > > >>>> spent a
> > > > > >>>>>>> lot of time to make Samza's KafkaSystemConsumer very
> robust.
> > It
> > > > > >>>>>>> also allows us to support some cool features:
> > > > > >>>>>>>
> > > > > >>>>>>> * Per-partition message ordering and prioritization.
> > > > > >>>>>>> * Tight control over partition assignment to support joins,
> > > > > >>>>>>> global
> > > > > >>>> state
> > > > > >>>>>>> (if we want to implement it :)), etc.
> > > > > >>>>>>> * Tight control over offset checkpointing.
> > > > > >>>>>>>
> > > > > >>>>>>> What we didn't realize at the time is that these features
> > > > > >>>>>>> should
> > > > > >>>>> actually
> > > > > >>>>>>> be in Kafka. A lot of Kafka consumers (not just Samza
> stream
> > > > > >>>> processors)
> > > > > >>>>>>> end up wanting to do things like joins and partition
> > > > > >>>>>>> assignment. The
> > > > > >>>>> Kafka
> > > > > >>>>>>> community has come to the same conclusion. They're adding a
> > ton
> > > > > >>>>>>> of upgrades into their new Kafka consumer implementation.
> To
> > a
> > > > > >>>>>>> large extent,
> > > > > >>> it's
> > > > > >>>>>>> duplicate work to what we've already done in Samza.
> > > > > >>>>>>>
> > > > > >>>>>>> On top of this, Kafka ended up taking a very similar
> approach
> > > > > >>>>>>> to
> > > > > >>>> Samza's
> > > > > >>>>>>> KafkaCheckpointManager implementation for handling offset
> > > > > >>>> checkpointing.
> > > > > >>>>>>> Like Samza, Kafka's new offset management feature stores
> > offset
> > > > > >>>>>>> checkpoints in a topic, and allows you to fetch them from
> the
> > > > > >>>>>>> broker.
> > > > > >>>>>>>
> > > > > >>>>>>> A lot of this seems like a waste, since we could have
> shared
> > > > > >>>>>>> the
> > > > > >>> work
> > > > > >>>> if
> > > > > >>>>>>> it
> > > > > >>>>>>> had been done in Kafka from the get-go.
> > > > > >>>>>>>
> > > > > >>>>>>> Vision
> > > > > >>>>>>>
> > > > > >>>>>>> All of this leads me to a rather radical proposal. Samza is
> > > > > >>> relatively
> > > > > >>>>>>> stable at this point. I'd venture to say that we're near a
> > 1.0
> > > > > >>>> release.
> > > > > >>>>>>> I'd
> > > > > >>>>>>> like to propose that we take what we've learned, and begin
> > > > > >>>>>>> thinking
> > > > > >>>>> about
> > > > > >>>>>>> Samza beyond 1.0. What would we change if we were starting
> > from
> > > > > >>>> scratch?
> > > > > >>>>>>> My
> > > > > >>>>>>> proposal is to:
> > > > > >>>>>>>
> > > > > >>>>>>> 1. Make Samza standalone the *only* way to run Samza
> > > > > >>>>>>> processors, and eliminate all direct dependences on YARN,
> > > Mesos,
> > > > > >> etc.
> > > > > >>>>>>> 2. Make a definitive call to support only Kafka as the
> stream
> > > > > >>>> processing
> > > > > >>>>>>> layer.
> > > > > >>>>>>> 3. Eliminate Samza's metrics, logging, serialization, and
> > > > > >>>>>>> config
> > > > > >>>>> systems,
> > > > > >>>>>>> and simply use Kafka's instead.
> > > > > >>>>>>>
> > > > > >>>>>>> This would fix all of the issues that I outlined above. It
> > > > > >>>>>>> should
> > > > > >>> also
> > > > > >>>>>>> shrink the Samza code base pretty dramatically. Supporting
> > only
> > > > > >>>>>>> a standalone container will allow Samza to be executed on
> > YARN
> > > > > >>>>>>> (using Slider), Mesos (using Marathon/Aurora), or most
> other
> > > > > >>>>>>> in-house
> > > > > >>>>> deployment
> > > > > >>>>>>> systems. This should make life a lot easier for new users.
> > > > > >>>>>>> Imagine
> > > > > >>>>> having
> > > > > >>>>>>> the hello-samza tutorial without YARN. The drop in mailing
> > list
> > > > > >>>> traffic
> > > > > >>>>>>> will be pretty dramatic.
> > > > > >>>>>>>
> > > > > >>>>>>> Coupling with Kafka seems long overdue to me. The reality
> is,
> > > > > >>> everyone
> > > > > >>>>>>> that
> > > > > >>>>>>> I'm aware of is using Samza with Kafka. We basically
> require
> > it
> > > > > >>>> already
> > > > > >>>>> in
> > > > > >>>>>>> order for most features to work. Those that are using other
> > > > > >>>>>>> systems
> > > > > >>>> are
> > > > > >>>>>>> generally using it for ingest into Kafka (1), and then they
> > do
> > > > > >>>>>>> the processing on top. There is already discussion (
> > > > > >>>
> > > >
> https://cwiki.apache.org/confluence/pages/viewpage.action?pageId=58851
> > > > > >>> 767
> > > > > >>>>>>> )
> > > > > >>>>>>> in Kafka to make ingesting into Kafka extremely easy.
> > > > > >>>>>>>
> > > > > >>>>>>> Once we make the call to couple with Kafka, we can
> leverage a
> > > > > >>>>>>> ton of
> > > > > >>>>> their
> > > > > >>>>>>> ecosystem. We no longer have to maintain our own config,
> > > > > >>>>>>> metrics,
> > > > > >>> etc.
> > > > > >>>>> We
> > > > > >>>>>>> can all share the same libraries, and make them better.
> This
> > > > > >>>>>>> will
> > > > > >>> also
> > > > > >>>>>>> allow us to share the consumer/producer APIs, and will let
> us
> > > > > >>> leverage
> > > > > >>>>>>> their offset management and partition management, rather
> than
> > > > > >>>>>>> having
> > > > > >>>> our
> > > > > >>>>>>> own. All of the coordinator stream code would go away, as
> > would
> > > > > >>>>>>> most
> > > > > >>>> of
> > > > > >>>>>>> the
> > > > > >>>>>>> YARN AppMaster code. We'd probably have to push some
> > partition
> > > > > >>>>> management
> > > > > >>>>>>> features into the Kafka broker, but they're already moving
> in
> > > > > >>>>>>> that direction with the new consumer API. The features we
> > have
> > > > > >>>>>>> for
> > > > > >>>> partition
> > > > > >>>>>>> assignment aren't unique to Samza, and seem like they
> should
> > be
> > > > > >>>>>>> in
> > > > > >>>> Kafka
> > > > > >>>>>>> anyway. There will always be some niche usages which will
> > > > > >>>>>>> require
> > > > > >>>> extra
> > > > > >>>>>>> care and hence full control over partition assignments much
> > > > > >>>>>>> like the
> > > > > >>>>> Kafka
> > > > > >>>>>>> low level consumer api. These would continue to be
> supported.
> > > > > >>>>>>>
> > > > > >>>>>>> These items will be good for the Samza community. They'll
> > make
> > > > > >>>>>>> Samza easier to use, and make it easier for developers to
> add
> > > > > >>>>>>> new features.
> > > > > >>>>>>>
> > > > > >>>>>>> Obviously this is a fairly large (and somewhat backwards
> > > > > >>> incompatible
> > > > > >>>>>>> change). If we choose to go this route, it's important that
> > we
> > > > > >>> openly
> > > > > >>>>>>> communicate how we're going to provide a migration path
> from
> > > > > >>>>>>> the
> > > > > >>>>> existing
> > > > > >>>>>>> APIs to the new ones (if we make incompatible changes). I
> > think
> > > > > >>>>>>> at a minimum, we'd probably need to provide a wrapper to
> > allow
> > > > > >>>>>>> existing StreamTask implementations to continue running on
> > the
> > > > > >> new container.
> > > > > >>>>> It's
> > > > > >>>>>>> also important that we openly communicate about timing, and
> > > > > >>>>>>> stages
> > > > > >>> of
> > > > > >>>>> the
> > > > > >>>>>>> migration.
> > > > > >>>>>>>
> > > > > >>>>>>> If you made it this far, I'm sure you have opinions. :)
> > Please
> > > > > >>>>>>> send
> > > > > >>>> your
> > > > > >>>>>>> thoughts and feedback.
> > > > > >>>>>>>
> > > > > >>>>>>> Cheers,
> > > > > >>>>>>> Chris
> > > > > >>>>
> > > > > >>>>
> > > > > >>>>
> > > > > >>>> --
> > > > > >>>> -- Guozhang
> > > > > >>
> > > > >
> > > >
> > >
> >
>
>
>
> --
> -- Guozhang
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message