Thanks for your reply @Fabian and @Stefan
@Fabian: The bloom filter state I proposal would be "elastic" and "lazy allocation", what we have on each key group is a list of bloom filter node(which is shrinkable), every bloom filter node has its capacity, we allocate a new one only when the last one is filled, and also there is a relaxed TTL to recycle the memory resources to enable it to run properly, for example, we consider the situation on key group 0: ------------------------------------------------------------------------------------ key group0: bfNode0 -> bfNode1 -> bfNode2 ------------------------------------------------------------------------------------ if the bfNode2 is filled, we allocate a new BF to store the data fails into key group 0, the situation on key group 0 becomes: ------------------------------------------------------------------------------------ key group0: bfNode0 -> bfNode1 -> bfNode2 -> bfNode3 ------------------------------------------------------------------------------------ and once a bfNode filled, it will never be changed again, so for example, if bfNode0 is filled at time point: t1, and the TTL = 2 hour, then we could release bfNode0 at the time point "t1 + 2hour". After that the situation on key group 0 becomes: ------------------------------------------------------------------------------------ key group0: bfNode1 -> bfNode2 -> bfNode3 ------------------------------------------------------------------------------------ What do you think of this? @Stefan: Yes, I definitely agree with your point. And we should discuss the implementation deeply before jump into implementation. Best, Sihua On 05/23/2018 17:33，Stefan Richter<s.rich...@data-artisans.com> wrote： Hi, In general, I like the proposal as well. We should try to integrate all forms of keyed state with the backend, to avoid the problems that we are currently facing with the timer service. We should discuss which exact implementation of bloom filters are the best fit. @Fabian: There are also implementations of bloom filters that use counting and therefore support deletes, but obviously this comes at the cost of a potentially higher space consumption. Best, Stefan Am 23.05.2018 um 11:29 schrieb Fabian Hueske <fhue...@gmail.com>: Thanks for the proposal Sihua! Let me try to summarize the motivation / scope of this proposal. You are proposing to add support for a special Bloom Filter state per KeyGroup and reduce the number of key accesses by checking the Bloom Filter first. This is would be a rather generic feature that could be interesting for various applications, including joins and deduplication as you described. IMO, such a feature would be very interesting. However, my concerns with Bloom Filter is that they are insert-only data structures, i.e., it is not possible to remove keys once they were added. This might render the filter useless over time. In a different thread (see discussion in FLINK-8918 [1]), you mentioned that the Bloom Filters would be growing. If we keep them in memory, how can we prevent them from exceeding memory boundaries over time? Best, Fabian [1] https://issues.apache.org/jira/browse/FLINK-8918 <https://issues.apache.org/jira/browse/FLINK-8918> 2018-05-23 9:56 GMT+02:00 sihua zhou <summerle...@163.com <mailto:summerle...@163.com>>: Hi Devs! I proposal to introduce "Elastic Bloom Filter" for Flink, the reason I make up this proposal is that, it helped us a lot on production, it let's improve the performance with reducing consumption of resources. Here is a brief description fo the motivation of why it's so powful, more detail information can be found https://issues.apache.org/jira/browse/FLINK-8601 <https://issues.apache.org/jira/browse/FLINK-8601> , and the design doc can be found https://docs.google.com/document/d/17UY5RZ1mq--hPzFx-LfBjCAw_kkoIrI9KHovXWkxNYY/edit?usp=sharing <https://docs.google.com/document/d/17UY5RZ1mq--hPzFx-LfBjCAw_kkoIrI9KHovXWkxNYY/edit?usp=sharing> ------------------------------------ Motivation There are some scenarios drive us to introduce this ElasticBloomFilter, one is Stream Join, another is Data Deduplication, and some special user cases...This has given us a great experience, for example, we implemented the Runtime Filter Join base on it, and it gives us a great performance improvement. With this feature, It diffs us from the "normal stream join", allows us to improve performance while reducing resource consumption by about half!!! I will list the two most typical user cases that optimized by the ElasticBloomFilter: one is "Runtime Filter Join" in detail, another is "Data Dedeplication" in brief. Scenario 1: Runtime Filter Join In general, stream join is one of the most performance cost task. For every record from both side, we need to query the state from the other side, this will lead to poor performance when the state size if huge. So, in production, we always need to spend a lot slots to handle stream join. But, indeed, we can improve this in somehow, there a phenomenon of stream join can be found in production. That's the “joined ratio” of the stream join is often very low, for example. stream join in promotion analysis: Job need to join the promotion log with the action(click, view, buy) log with the promotion_id to analysis the effect of the promotion. stream join in AD(advertising) attribution: Job need to join the AD click log with the item payment log on the click_id to find which click of which AD that brings the payment to do attribution. stream join in click log analysis of doc: Job need to join viewed log(doc viewed by users) with the click log (doc clicked by users) to analysis the reason of the click and the property of the users. ….so on All these cases have one common property, that is the joined ratio is very low. Here is a example to describe it, we have 10000 records from the left stream, and 10000 records from the right stream, and we execute select * from leftStream l join rightStream r on l.id <http://l.id/> = r.id <http://r.id/> , we only got 100 record from the result, that is the case for low joined ratio, this is an example for inner join, but it can also applied to left & right join. there are more example I can come up with low joined ratio…but the point I want to raise up is that the low joined ratio of stream join in production is a very common phenomenon(maybe even the almost common phenomenon in some companies, at least in our company that is the case). How to improve this? We can see from the above case, 10000 record join 10000 record and we only got 100 result, that means, we query the state 20000 times (10000 for the left stream and 10000 for the right stream) but only 100 of them are meaningful!!! If we could reduce the useless query times, then we can definitely improve the performance of stream join. the way we used to improve this is to introduce the Runtime Filter Join, the mainly ideal is that, we build a filter for the state on each side (left stream & right stream). When we need to query the state on that side we first check the corresponding filter whether the key is possible in the state, if the filter say "not, it impossible in the State", then we stop querying the state, if it say "hmm, it maybe in state", then we need to query the state. As you can see, the best choose of the filter is Bloom Filter, it has all the feature that we want: extremely good performance, non-existence of false negative. Scenario 2: Data Deduplication We have implemented two general functions based on the ElasticBloomFilter. They are count(distinct x) and select distinct x, y, z from table. Unlike the Runtime Filter Join the result of this two functions is approximate, not exactly. There are used in the scenario where we don't need a 100% accurate result, for example, to count the number of visiting users in each online store. In general, we don't need a 100% accurate result in this case(indeed we can't give a 100% accurate result, because there could be error when collecting user_id from different devices), if we could get a 98% accurate result with only 1/2 resource, that could be very nice. I believe there would be more user cases in stream world that could be optimized by the Bloom Filter(as what it had done in the big data world)... I will appreciate it very much, if someone could have a look of the JIRA or the google doc and give some comments! Thanks, Sihua
