DynamoDB concurrent write, result in throttling - database

Below is my structure for table
table
UUID- key - Let call this **EntryKey**
HistoryLog - this also version number
Map<UUID (Let call this **EntryChildKey**, BYTE> value
version - For **optimistic locking**
Let's assume map has around 10k entry uuid to some value.
So, my problem is once in while I am getting request to update 10k EntryChildKey(map) value and all this request bombard db at the same time and because, every time I am hitting same EntryKey row, I am running in to lot of concurrency error, version got update every time and I have to retry and all EntryChildKey updates are thrashing each other, resulting in DynamoDB throttling my request.
I can get out of this problem if I separate this in to 2 tables as below, but we have to maintain HistoryLog version changing at EntryKey level and also there are some other problem so I can’t take this route
Table1 Table2
UUID EntryChildKey UUID EntryKey
BYTE value List<UUID> EntryChildKey
So, another approach I am thinking is Write ahead log kind of stuff, where I’ll update the version and also record the intent to update the table, but won’t update the record, instead keep it as list in table and then update the EntryChildKey values sequentially. But, I don’t whether there is something like this or similar thing I can do with DynamoDb or not ?
Also any another approach that could help to solve this problem I’ll appreciate

If you really do need to have a version attribute be updated on a single key each time any one of the 10k EntryChild items is updated then your only option is to decouple the table from the update source.
DynamoDB has a hard limit of up to 1000 writes/second to any item at all times. There is simply no way to increase that, for a single item. It doesn't matter what size table you have, how many partitions, or how much total write capacity you allocate to your table, a single item will never be able to be updated more than 1000 times per second.
So, if your requirement to update an attribute (the HistoryLog in your example) on the "master" entry item is really firm, then to use DynamoDB your best bet is to introduce a queue and batching to pre-process the updates before writing to Dynamo.
You could create an SQS queue and use a lambda function to read from the queue and write to Dynamo.
In a naive approach, you could simple read from the queue and then write to the table as much as you can, based on the DynamoDB throttling. For 10k updates to the same "master" key this will take at least 10 seconds, though in reality it will likely take longer.
A better option though, would be to run the lambda on a schedule, say once a second, and have it read all the messages available in the queue and combine all updates to the same "master" key into a single update. That way, you only write to the same item at most once every second.
The big challenge with a normal SQS queue is that it does not offer exactly once semantics: meaning there will be items in the queue that will be received multiple times. If you can design a system where you can safely discard duplicate updates then this approach will work wonderful. If not, then things get more complicated.

Related

Updating database keys where one table's keys refer to another's

I have two tables in DynamoDB. One has data about homes, one has data about businesses. The homes table has a list of the closest businesses to it, with walking times to each of them. That is, the homes table has a list of IDs which refer to items in the businesses table. Since businesses are constantly opening and closing, both these tables need to be updated frequently.
The problem I'm facing is that, when either one of the tables is updated, the other table will have incorrect data until it is updated itself. To make this clearer: let's say one business closes and another one opens. I could update the businesses table first to remove the old business and add the new one, but the homes table would then still refer to the now-removed business. Similarly, if I updated the homes table first to refer to the new business, the businesses table would not yet have this new business' data yet. Whichever table I update first, there will always be a period of time where the two tables are not in synch.
What's the best way to deal with this problem? One way I've considered is to do all the updates to a secondary database and then swap it with my primary database, but I'm wondering if there's a better way.
Thanks!
Dynamo only offers atomic operations on the item level, not transaction level, but you can have something similar to an atomic transaction by enforcing some rules in your application.
Let's say you need to run a transaction with two operations:
Delete Business(id=123) from the table.
Update Home(id=456) to remove association with Business(id=123) from the home.businesses array.
Here's what you can do to mimic a transaction:
Generate a timestamp for locking the items
Let's say our current timestamp is 1234567890. Using a timestamp will allow you to clean up failed transactions (I'll explain later).
Lock the two items
Update both Business-123 and Home-456 and set an attribute lock=1234567890.
Do not change any other attributes yet on this update operation!
Use a ConditionalExpression (check the Developer Guide and API) to verify that attribute_not_exists(lock) before updating. This way you're sure there's no other process using the same items.
Handle update lock responses
Check if both updates succeeded to Home and Business. If yes to both, it means you can proceed with the actual changes you need to make: delete the Business-123 and update the Home-456 removing the Business association.
For extra care, also use a ConditionExpression in both updates again, but now ensuring that lock == 1234567890. This way you're extra sure no other process overwrote your lock.
If both updates succeed again, you can consider the two items updated and consistent to be read by other processes. To do this, run a third update removing the lock attribute from both items.
When one of the operations fail, you may try again X times for example. If it fails all X times, make sure the process cleans up the other lock that succeeded previously.
Enforce the transaction lock throught your code
Always use a ConditionExpression in any part of your code that may update/delete Home and Business items. This is crucial for the solution to work.
When reading Home and Business items, you'll need to do this (this may not be necessary in all reads, you'll decide if you need to ensure consistency from start to finish while working with an item read from DB):
Retrieve the item you want to read
Generate a lock timestamp
Update the item with lock=timestamp using a ConditionExpression
If the update succeeds, continue using the item normally; if not, wait one or two seconds and try again;
When you're done, update the item removing the lock
Regularly clean up failed transactions
Every minute or so, run a background process to look for potentially failed transactions. If your processes take at max 60 seconds to finish and there's an item with lock value older than, say 5 minutes (remember lock value is the time the transaction started), it's safe to say that this transaction failed at some point and whatever process running it didn't properly clean up the locks.
This background job would ensure that no items keep locked for eternity.
Beware this implementation do not assure a real atomic and consistent transaction in the sense traditional ACID DBs do. If this is mission critical for you (e.g. you're dealing with financial transactions), do not attempt to implement this. Since you said you're ok if atomicity is broken on rare failure occasions, you may live with it happily. ;)
Hope this helps!

how to use Google Cloud Memcach to save/update unique items

In my application I run a cron job to loop over all users (2500 user) to choose an item for every user out of 4k items, considering that:
- choosing the item is based on some user info,
- I need to make sure that each user take a unique item that wasn't taken by any one else, so relation is one-to-one
To achieve this I have to run this cron job and loop over the users one by one sequentially and pick up the item for each then remove it from the list (not to be chosen by next user(s)) then move to the next user
actually in my system the number of users/items is getting bigger and bigger every single day, this cron job now takes 2 hours to set items to all users.
I need to improve this, one of the things I've thought about is using Threads but I cant do that since Im using automatic scaling, so I start thinking about push Queues, so when the cron jobs run, will make a loop like this:
for(User user : users){
getMyItem(user.getId());
}
where getMyItem will push the task to a servlet to handle it and choose the best item for this person based on his data.
Let's say I'll start doing that so what will be the best/robust solution to avoid setting an item to more than one user ?
Since Im using basic scaling and 8 instances, can't rely on static variables.
one of the things that came across my mind is to create a table in the DB that accept only unique items then I insert into it the taken items so if the insertion is done successfully it means no body else took this item so i can just assign it to that person, but this will make the performance a bit lower cause I need to make write DB operation with every call (I want to avoid that)
Also I thought about MemCach, its really fast but not robust enough, if I save a Set of items into it which will accept only unique items, then if more than one thread was trying to access this Set at the same time to update it, only one thread will be able to save its data and all other threads data might be overwritten and lost.
I hope you guys can help to find a solution for this problem, thanks in advance :)
First - I would advice against using solely memcache for such algorithm - the key thing to remember about memcache is that it is volatile and might dissapear at any time, breaking the algorithm.
From Service levels:
Note: Whether shared or dedicated, memcache is not durable storage. Keys can be evicted when the cache fills up, according to the
cache's LRU policy. Changes in the cache configuration or datacenter
maintenance events can also flush some or all of the cache.
And from How cached data expires:
Under rare circumstances, values can also disappear from the cache
prior to expiration for reasons other than memory pressure. While
memcache is resilient to server failures, memcache values are not
saved to disk, so a service failure can cause values to become
unavailable.
I'd suggest adding a property, let's say called assigned, to the item entities, by default unset (or set to null/None) and, when it's assigned to a user, set to the user's key or key ID. This allows you:
to query for unassigned items when you want to make assignments
to skip items recently assigned but still showing up in the query results due to eventual consistency, so no need to struggle for consistency
to be certain that an item can uniquely be assigned to only a single user
to easily find items assigned to a certain user if/when you're doing per-user processing of items, eventually setting the assigned property to a known value signifying done when its processing completes
Note: you may need a one-time migration task to update this assigned property for any existing entities when you first deploy the solution, to have these entities included in the query index, otherwise they would not show up in the query results.
As for the growing execution time of the cron jobs: just split the work into multiple fixed-size batches (as many as needed) to be performed in separate requests, typically push tasks. The usual approach for splitting is using query cursors. The cron job would only trigger enqueueing the initial batch processing task, which would then enqueue an additional such task if there are remaining batches for processing.
To get a general idea of such a solution works take a peek at Google appengine: Task queue performance (it's python, but the general idea is the same).
If you are planning for push jobs inside a cron and you want the jobs to be updating key-value pairs as an addon to improvise the speed and performance, we can split the number of users and number of items into multiple key-(list of values) pairs so that our push jobs will pick the key random ( logic to write to pick a key out of 4 or 5 keys) and then remove an item from the list of items and update the key again, try to have a locking before working on the above part. Example of key value paris.
Userlist1: ["vijay",...]
Userlist2: ["ramana",...]

Paging of frequently changing data

I'm developing a web application which display a list of let's say "threads". The list can be sorted by the amount of likes a thread has. There can be thousands of threads in one list.
The application needs to work in a scenario where the likes of a thread can change more than 10x in a second. The application furthermore is distributed over multiple servers.
I can't figure out an efficient way to enable paging for this sort of list. And I can't transmit the whole sorted list by likes to a user at once.
As soon as an user would go to page 2 of this list, it likely changed and may contain threads already listed from page one
Solutions which don't work:
Storing the seen threads on the client side (could be too many on mobile)
Storing the seen threads on the Server side (too many users and threads)
Snapshot the list in temp database table (it's too frequent changing data and it need to be actual)
(If it matters I'm using MongoDB+c#)
How would you solve this kind of problem?
Interesting question. Unless I'm misunderstanding you, and by all means let me know if I am, it sounds like the best solution would be to implement a system that, instead of page numbers, uses timestamps. It would be similar to what many of the main APIs already do. I know Tumblr even does this on the dashboard, where this is, of course, not an unreasonable case: there can be tons of posts added in a small amount of time at peak hours, depending on how many people the user follows.
So basically, your "next page" button could just link to /threads/threadindex/1407051000, which could translate to "all the threads that were created before 2014-08-02 17:30. That makes your query super easy to implement. Then, when you pull down all the next elements, you just look for anything that occurred before the last element on the page.
The downfall of this, of course, is that it's hard to know how many new elements have been added since the user started browsing, but you could always log the start time and know anything since then would be new. And it's also difficult for users to type in their own pages, but that's not a problem in most applications. You also need to store the timestamps for every record in your thread, but that's probably already being done, and if it's not then it's certainly not hard to implement. You'll be paying the cost of something like eight bytes extra per record, but that's better than having to store anything about "seen" posts.
It's also nice because, and again this might not apply to you, but a user could bookmark a page in the list, and it would last unchanged forever since it's not relative to anything else.
This is typically handled using an OLAP cube. The idea here is that you add a natural time dimension. They may be too heavy for this application, but here's a summary in case someone else needs it.
OLAP cubes start with the fundamental concept of time. You have to know what time you care about to be able to make sense of the data.
You start off with a "Time" table:
Time {
timestamp long (PK)
created datetime
last_queried datetime
}
This basically tracks snapshots of your data. I've included a last_queried field. This should be updated with the current time any time a user asks for data based on this specific timestamp.
Now we can start talking about "Threads":
Threads {
id long (PK)
identifier long
last_modified datetime
title string
body string
score int
}
The id field is an auto-incrementing key; this is never exposed. identifier is the "unique" id for your thread. I say "unique" because there's no unique-ness constraint, and as far as the database is concerned it is not unique. Everything else in there is pretty standard... except... when you do writes you do not update this entry. In OLAP cubes you almost never modify data. Updates and inserts are explained at the end.
Now, how do we query this? You can't just directly query Threads. You need to include a star table:
ThreadStar {
timestamp long (FK -> Time.timestamp)
thread_id long (FK -> Threads.id)
thread_identifier long (matches Threads[thread_id].identifier)
(timestamp, thread_identifier should be unique)
}
This table gives you a mapping from what time it is to what the state of all of the threads are. Given a specific timestamp you can get the state of a Thread by doing:
SELECT Thread.*
FROM Thread
JOIN ThreadStar ON Thread.id = ThreadStar.thread_id
WHERE ThreadStar.timestamp = {timestamp}
AND Thread.identifier = {thread_identifier}
That's not too bad. How do we get a stream of threads? First we need to know what time it is. Basically you want to get the largest timestamp from Time and update Time.last_queried to the current time. You can throw a cache up in front of that that only updates every few seconds, or whatever you want. Once you have that you can get all threads:
SELECT Thread.*
FROM Thread
JOIN ThreadStar ON Thread.id = ThreadStar.thread_id
WHERE ThreadStar.timestamp = {timestamp}
ORDER BY Thread.score DESC
Nice. We've got a list of threads and the ordering is stable as the actual scores change. You can page through this at your leisure... kind of. Eventually data will be cleaned up and you'll lose your snapshot.
So this is great and all, but now you need to create or update a Thread. Creation and modification are almost identical. Both are handled with an INSERT, the only difference is whether you use an existing identifier or create a new one.
So now you've inserted a new Thread. You need to update ThreadStar. This is the crazy expensive part. Basically you make a copy of all of the ThreadStar entries with the most recent timestamp, except you update the thread_id for the Thread you just modified. That's a crazy amount of duplication. Fortunately it's pretty much only foreign keys, but still.
You also don't do DELETEs either; mark a row as deleted or just exclude it when you update ThreadStar.
Now you're humming along, but you've got crazy amounts of data growing. You'll probably want to clean it out, unless you've got a lot of storage budge, but even then things will start slowing down (aside: this will actually perform shockingly well, even with crazy amounts of data).
Cleanup is pretty straightforward. It's just a matter of some cascading deletes and scrubbing for orphaned data. Delete entries from Time whenever you want (e.g. it's not the latest entry and last_queried is null or older than whatever cutoff). Cascade those deletes to ThreadStar. Then find any Threads with an id that isn't in ThreadStar and scrub those.
This general mechanism also works if you have more nested data, but your queries get harder.
Final note: you'll find that your inserts get really slow because of the sheer amounts of data. Most places build this with appropriate constraints in development and testing environments, but then disable constraints in production!
Yeah. Make sure your tests are solid.
But at least you aren't sensitive to re-ordered data mid-paging.
For constantly changing data such as likes I would use a two stage appraoch. For the frequently changing data I would use an in memory DB to keep up with the change rates and flush this peridically to the "real" db.
Once you have that the query for constantly chaning data is easy.
Query the db.
Query the in memory db.
Merge the frequently changed data from the in memory db with the "slow" db data .
Remember which results you already have displayed so pressing the next button will
not display an already dispalyed value twice because on different pages because its rank has changed.
If many people look at the same data it might help to cache the results of 3 in itself to reduce the load on the real db even further.
Your current architecture has no caching layers (the bigger the site the more things are cached). You will not get away with a simple DB and efficient queries against the db if things become too massive.
I would cache all 'thread' results on the server when the user first time hits the database. Then return the first page of data to the user and for each subsequent next page calls I'd return cached results.
To minimize memory usage you can cache only records ids and fetch whole data when user requests it.
Cache can be evicted each time user exits current page. If it isn't a ton of data I would stick to this solution because user won't get annoyed of data constantly changing.

app engine data pipelines talk - for fan-in materialized view, why are work indexes necessary?

I'm trying to understand the data pipelines talk presented at google i/o:
http://www.youtube.com/watch?v=zSDC_TU7rtc
I don't see why fan-in work indexes are necessary if i'm just going to batch through input-sequence markers.
Can't the optimistically-enqueued task grab all unapplied markers, churn through as many of them as possible (repeatedly fetching a batch of say 10, then transactionally update the materialized view entity), and re-enqueue itself if the task times out before working through all markers?
Does the work indexes have something to do with the efficiency querying for all unapplied markers? i.e., it's better to query for "markers with work_index = " than for "markers with applied = False"? If so, why is that?
For reference, the question+answer which led me to the data pipelines talk is here:
app engine datastore: model for progressively updated terrain height map
A few things:
My approach assumes multiple workers (see ShardedForkJoinQueue here: http://code.google.com/p/pubsubhubbub/source/browse/trunk/hub/fork_join_queue.py), where the inbound rate of tasks exceeds the amount of work a single thread can do. With that in mind, how would you use a simple "applied = False" to split work across N threads? Probably assign another field on your model to a worker's shard_number at random; then your query would be on "shard_number=N AND applied=False" (requiring another composite index). Okay that should work.
But then how do you know how many worker shards/threads you need? With the approach above you need to statically configure them so your shard_number parameter is between 1 and N. You can only have one thread querying for each shard_number at a time or else you have contention. I want the system to figure out the shard/thread count at runtime. My approach batches work together into reasonably sized chunks (like the 10 items) and then enqueues a continuation task to take care of the rest. Using query cursors I know that each continuation will not overlap the last thread's, so there's no contention. This gives me a dynamic number of threads working in parallel on the same shard's work items.
Now say your queue backs up. How do you ensure the oldest work items are processed first? Put another way: How do you prevent starvation? You could assign another field on your model to the time of insertion-- call it add_time. Now your query would be "shard_number=N AND applied=False ORDER BY add_time DESC". This works fine for low throughput queues.
What if your work item write-rate goes up a ton? You're going to be writing many, many rows with roughly the same add_time. This requires a Bigtable row prefix for your entities as something like "shard_number=1|applied=False|add_time=2010-06-24T9:15:22". That means every work item insert is hitting the same Bigtable tablet server, the server that's currently owner of the lexical head of the descending index. So fundamentally you're limited to the throughput of a single machine for each work shard's Datastore writes.
With my approach, your only Bigtable index row is prefixed by the hash of the incrementing work sequence number. This work_index value is scattered across the lexical rowspace of Bigtable each time the sequence number is incremented. Thus, each sequential work item enqueue will likely go to a different tablet server (given enough data), spreading the load of my queue beyond a single machine. With this approach the write-rate should effectively be bound only by the number of physical Bigtable machines in a cluster.
One disadvantage of this approach is that it requires an extra write: you have to flip the flag on the original marker entity when you've completed the update, which is something Brett's original approach doesn't require.
You still need some sort of work index, too, or you encounter the race conditions Brett talked about, where the task that should apply an update runs before the update transaction has committed. In your system, the update would still get applied - but it could be an arbitrary amount of time before the next update runs and applies it.
Still, I'm not the expert on this (yet ;). I've forwarded your question to Brett, and I'll let you know what he says - I'm curious as to his answer, too!

Cost of Inserts vs Update in SQL Server

I have a table with more than a millon rows. This table is used to index tiff images. Each image has fields like date, number, etc. I have users that index these images in batches of 500. I need to know if it is better to first insert 500 rows and then perform 500 updates or, when the user finishes indexing, to do the 500 inserts with all the data. A very important thing is that if I do the 500 inserts at first, this time is free for me because I can do it the night before.
So the question is: is it better to do inserts or inserts and updates, and why? I have defined a id value for each image, and I also have other indices on the fields.
Updates in Sql server result in ghosted rows - i.e. Sql crosses one row out and puts a new one in. The crossed out row is deleted later.
Both inserts and updates can cause page-splits in this way, they both effectively 'add' data, it's just that updates flag the old stuff out first.
On top of this updates need to look up the row first, which for lots of data can take longer than the update.
Inserts will just about always be quicker, especially if they are either in order or if the underlying table doesn't have a clustered index.
When inserting larger amounts of data into a table look at the current indexes - they can take a while to change and build. Adding values in the middle of an index is always slower.
You can think of it like appending to an address book: Mr Z can just be added to the last page, while you'll have to find space in the middle for Mr M.
Doing the inserts first and then the updates does seem to be a better idea for several reasons. You will be inserting at a time of low transaction volume. Since inserts have more data, this is a better time to do it.
Since you are using an id value (which is presumably indexed) for updates, the overhead of updates will be very low. You would also have less data during your updates.
You could also turn off transactions at the batch (500 inserts/updates) level and use it for each individual record, thus reducing some overhead.
Finally, test this out to see the actual performance on your server before making a final decision.
This isn't a cut and dry question. Krishna's and Galegian's points are spot on.
For updates, the impact will be lessened if the updates are affecting fixed-length fields. If updating varchar or blob fields, you may add a cost of page splits during update when the new value surpasses the length of the old value.
I think inserts will run faster. They do not require a lookup (when you do an update you are basically doing the equivalent of a select with the where clause). And also, an insert won't lock the rows the way an update will, so it won't interfere with any selects that are happening against the table at the same time.
The execution plan for each query will tell you which one should be more expensive. The real limiting factor will be the writes to disk, so you may need to run some tests while running perfmon to see which query causes more writes and causes the disk queue to get the longest (longer is bad).
I'm not a database guy, but I imagine doing the inserts in one shot would be faster because the updates require a lookup whereas the inserts do not.

Resources