I'm working on a system that mirrors remote datasets using initials and deltas. When an initial comes in, it mass deletes anything preexisting and mass inserts the fresh data. When a delta comes in, the system does a bunch of work to translate it into updates, inserts, and deletes. Initials and deltas are processed inside long transactions to maintain data integrity.
Unfortunately the current solution isn't scaling very well. The transactions are so large and long running that our RDBMS bogs down with various contention problems. Also, there isn't a good audit trail for how the deltas are applied, making it difficult to troubleshoot issues causing the local and remote versions of the dataset to get out of sync.
One idea is to not run the initials and deltas in transactions at all, and instead to attach a version number to each record indicating which delta or initial it came from. Once an initial or delta is successfully loaded, the application can be alerted that a new version of the dataset is available.
This just leaves the issue of how exactly to compose a view of a dataset up to a given version from the initial and deltas. (Apple's TimeMachine does something similar, using hard links on the file system to create "view" of a certain point in time.)
Does anyone have experience solving this kind of problem or implementing this particular solution?
Thanks!
have one writer and several reader databases. You send the write to the one database, and have it propagate the exact same changes to all the other databases. The reader databases will be eventually consistent and the time to update is very fast. I have seen this done in environments that get upwards of 1M page views per day. It is very scalable. You can even put a hardware router in front of all the read databases to load balance them.
Thanks to those who tried.
For anyone else who ends up here, I'm benchmarking a solution that adds a "dataset_version_id" and "dataset_version_verb" column to each table in question. A correlated subquery inside a stored procedure is then used to retrieve the current dataset_version_id when retrieving specific records. If the latest version of the record has dataset_version_verb of "delete", it's filtered out of the results by a WHERE clause.
This approach has an average ~ 80% performance hit so far, which may be acceptable for our purposes.
Related
I am in the middle of an interview simulation and I got stock with one question. Can someone provide the answer for me please?
The question:
We use a secondary datastore (we use elasticsearch alongside our main database) for real time analytics and reporting. What problems might you anticipate with this sort of approach? Explain how would go about solving or mitigating them?
Thank you
There are several problems:
No transactional cover : If your main database is transactional (which it usually is), so you either commit or you don't. After the record is inserted into your main database, there is no guarentee that it will be committed to ES. In fact if you commit several records to your primary DB, you may have a situation where some of them are committed to ES, and few others are not. This is a MAJOR issue.
Refresh Interval : Elasticsearch by default refreshes every second. That means "Real-time" is generally 1 second later, or at least when the data is queried for. If you commit a record into your primary db, and immediately query for it via ES, it may not get found. THe only way around this is to GET the record using its ID.
Data-Duplication : Elasticsearch cannot do joins. You need to denormalize all data that is coming from a RDBMS. If one user has many posts, you cannot "join" to search. You have to add the user id an any other user specific details to every post object.
Hardware : Elasticsearch needs RAM (bare minimum of 1 gb) to work properly. This is assuming you don't use anything else from the ELK stack. THis is an important cost wise consideration.
One problem might be synchronization issues, where the elastic search store gets out of sync and starts service stale data. To avoid issues, you will have to implement monitoring on your data pipeline, elastic search and the primary database, to detect any problem by checking for update times, delay, number of records (within some level of error) in each of them and overall system operation status (up / down).
Another is disconnection and recovery - what happens if your data pipeline or elastic search loses connection to the rest of the system? You will need an automatic way to re-connect, when network is restored and start synchronising data again.
You also have to take into account sudden influx of data - how to scale ElasticSearch ingestion or your data processor (data pipeline) if there is large amount of updates and inserts in peak hours or after re-connection when there was network issues.
I have previously done some very basic real-time applications using the help of sockets and have been reading more about it just for curiosity. One very interesting article I read was about Operational Transformation and I learned several new things. After reading it, I kept thinking of when or how this data is really saved to the database if I were to keep it. I have two assumptions/theories about what might be going on, but I'm not sure if they are correct and/or the best solutions to solve this issue. They are as follow:
(For this example lets assume it's a real-time collaborative whiteboard:)
For every edit that happens (ex. drawing a line), the socket will send a message to everyone collaborating. But at the same time, I will store the data in my database. The problem I see with this solution is the amount of time I would need to access the database. For every line a user draws, I would be required to access the database to store it.
Use polling. For this theory, I think of saving every data in temporal storage at the server, and then after 'x' amount of time, it will get all the data from the temporal storage and save them in the database. The issue for this theory is the possibility of a failure in the temporal storage (ex. electrical failure). If the temporal storage loses its data before it is saved in the database, then I would never be able to recover them again.
How do similar real-time collaborative applications like Google Doc, Slides, etc stores the data in their databases? Are they following one of the theories I mentioned or do they have a completely different way to store the data?
They prolly rely on logs of changes + latest document version + periodic snapshot (if they allow time traveling the document history).
It is similar to how most database's transaction system work. After validation the change is legit, the database writes the change in very fast data-structure on disk aka. the log that will only append the changed values. This log is replicated in-memory with a dedicated data-structure to speed up reads.
When a read comes in, the database will check the in-memory data-structure and merge the change with what is stored in the cache or on the disk.
Periodically, the changes that are present in memory and in the log, are merged with the data-structure on-disk.
So to summarize, in your case:
When an Operational Transformation comes to the server, two things happens:
It is stored in the database as is, to avoid any loss (equivalent of the log)
It updates an in-memory datastructure to be able to replay the change quickly in case an user request the latest version (equivalent of the memory datastructure)
When an user request the latest document, the server check the in-memory datastructre and replay the changes against the last stored consolidated document that might be lagging behind because of the following point
Periodically, the log is applied to the "last stored consolidated document" to reduce the amount of OT that must be replayed to produce the latest document.
Anyway, the best way to have a definitive answer is to look at open-source code that does what you are looking for, e.g. etherpad.
I currently have trouble updating SQLite database records at scale within a healthy amount of time.
I have a small database of about 70,000 records and I have some office personal using Navicat to filter the records and make some bulk edits to commit. When trying to perform an UPDATE on a large amount of records in one field everything comes to a crawl, when I look at the raw SQL query I can see that the program is using an UPDATE ..SET.. WHERE to perform the operations.
My question is what can I do to help this query run faster? I have a SQLite auto index on the column used for matching the record to update, but I have read and searched all over and have yet to see any kind of resolve other then what I already have in place. Updating one field in 20,000 records is literally taking close to 3 hours...regardless of using Navicat or not. The whole database is only 30mb so I have to be doing something wrong.
All other database operations run nice and speedy, not sure whats wrong and looking for some guidance from a SQLite vet.
There are very few reasons why SQLite can perform as poorly as you describe, but I've experienced pretty much the same thing.
First, if you are performing a table-wide update on an indexed column, you're gonna thrash the whole index, all while making the index less useless. Delete the index before you update, then recreate the index.
Additionally, you can get some serious perf boost by changing the synchronous PRAGMA, but it comes with a non-zero amount of extra risk of corrupt data on a crash (extremely unlikely in my experience)
PRAGMA schema.synchronous = 0
I have a problem where I need to load alot of data (5+ billion rows) into a database very quickly (ideally less than an 30 min but quicker is better), and I was recently suggested to look into postgresql (I failed with mysql and was looking at hbase/cassandra). My setup is I have a cluster (currently 8 servers) that generates alot of data, and I was thinking of running databases locally on each machine in the cluster it writes quickly locally and then at the end (or throughout the data generating) data is merged together. The data is not in any order so I don't care which specific server its on (as long as its eventually there).
My questions are , is there any good tutorials or places to learn about PostgreSQL auto sharding (I found results of firms like sykpe doing auto sharding but no tutorials, I want to play with this myself)? Is what I'm trying to do possible? Because the data is not in any order I was going to use auto-incrementing ID number, will that cause a conflict if data is merged (this is not a big issue anymore)?
Update: Frank's idea below kind of eliminated the auto-incrementing conflict issue I was asking about. The question is basically now, how can I learn about auto sharding and would it support distributed uploads of data to multiple servers?
First: Do you really need to insert the generated data from your cluster straight into a relational database? You don't mind merging it at the end anyway, so why bother inserting into a database at all? In your position I'd have your cluster nodes write flat files, probably gzip'd CSV data. I'd then bulk import and merge that data using a tool like pg_bulkload.
If you do need to insert directly into a relational database: That's (part of) what PgPool-II and (especeially) PgBouncer are for. Configure PgBouncer to load-balance across different nodes and you should be pretty much sorted.
Note that PostgreSQL is a transactional database with strong data durability guarantees. That also means that if you use it in a simplistic way, doing lots of small writes can be slow. You have to consider what trade-offs you're willing to make between data durability, speed, and cost of hardware.
At one extreme, each INSERT can be its own transaction that's synchronously committed to disk before returning success. This limits the number of transactions per second to the number of fsync()s your disk subsystem can do, which is often only in the tens or hundreds per second (without battery backup RAID controller). This is the default if you do nothing special and if you don't wrap your INSERTs in a BEGIN and COMMIT.
At the other extreme, you say "I really don't care if I lose all this data" and use unlogged tables for your inserts. This basically gives the database permission to throw your data away if it can't guarantee it's OK - say, after an OS crash, database crash, power loss, etc.
The middle ground is where you will probably want to be. This involves some combination of asynchronous commit, group commits (commit_delay and commit_siblings), batching inserts into groups wrapped in explicit BEGIN and END, etc. Instead of INSERT batching you could do COPY loads of a few thousand records at a time. All these things trade data durability off against speed.
For fast bulk inserts you should also consider inserting into tables without any indexes except a primary key. Maybe not even that. Create the indexes once your bulk inserts are done. This will be a hell of a lot faster.
Here are a few things that might help:
The DB on each server should have a small meta data table with that server's unique characteristics. Such as which server it is; servers can be numbered sequentially. Apart from the contents of that table, it's probably wise to try to keep the schema on each server as similar as possible.
With billions of rows you'll want bigint ids (or UUID or the like). With bigints, you could allocate a generous range for each server, and set its sequence up to use it. E.g. server 1 gets 1..1000000000000000, server 2 gets 1000000000000001 to 2000000000000000 etc.
If the data is simple data points (like a temperature reading from exactly 10 instruments every second) you might get efficiency gains by storing it in a table with columns (time timestamp, values double precision[]) rather than the more correct (time timestamp, instrument_id int, value double precision). This is an explicit denormalisation in aid of efficiency. (I blogged about my own experience with this scheme.)
Use citus for PostgreSQL auto sharding. Also this link is helpful.
Sorry I don't have a tutorial at hand, but here's an outline of a possible solution:
Load one eight of your data into a PG instance on each of the servers
For optimum load speed, don't use inserts but the COPY method
When the data is loaded, do not combine the eight databases into one. Instead, use plProxy to launch a single statement to query all databases at once (or the right one to satisfy your query)
As already noted, keys might be an issue. Use non-overlapping sequences or uuids or sequence numbers with a string prefix, shouldn't be too hard to solve.
You should start with a COPY test on one of the servers and see how close to your 30-minute goal you can get. If your data is not important and you have a recent Postgresql version, you can try using unlogged tables which should be a lot faster (but not crash-safe). Sounds like a fun project, good luck.
You could use mySQL - which supports auto-sharding across a cluster.
Where I'm at there is a main system that runs on a big AIX mainframe. To facility reporting and operations there is nightly dump from the mainframe into SQL Server, such that each of our 50-ish clients is in their own database with identical schemas. This dump takes about 7 hours to finish each night, and there's not really anything we can do about it: we're stuck with what the application vendor has provided.
After the dump into sql server we use that to run a number of other daily procedures. One of those procedures is to import data into a kind of management reporting sandbox table, which combines records from a particularly important table from across the different databases into one table that managers who don't know sql so can use to run ad-hoc reports without hosing up the rest of the system. This, again, is a business thing: the managers want it, and they have the power to see that we implement it.
The import process for this table takes a couple hours on it's own. It filters down about 40 million records spread across 50 databases into about 4 million records, and then indexes them on certain columns for searching. Even at a coupld hours it's still less than a third as long as the initial load, but we're running out of time for overnight processing, we don't control the mainframe dump, and we do control this. So I've been tasked with looking for ways to improve one the existing procedure.
Currently, the philosophy is that it's faster to load all the data from each client database and then index it afterwards in one step. Also, in the interest of avoiding bogging down other important systems in case it runs long, a couple of the larger clients are set to always run first (the main index on the table is by a clientid field). One other thing we're starting to do is load data from a few clients at a time in parallel, rather than each client sequentially.
So my question is, what would be the most efficient way to load this table? Are we right in thinking that indexing later is better? Or should we create the indexes before importing data? Should we be loading the table in index order, to avoid massive re-ordering of pages, rather than the big clients first? Could loading in parallel make things worse by causing to much disk access all at once or removing our ability to control the order? Any other ideas?
Update
Well, something is up. I was able to do some benchmarking during the day, and there is no difference at all in the load time whether the indexes are created at the beginning or at the end of the operation, but we save the time building the index itself (it of course builds nearly instantly with no data in the table).
I have worked with loading bulk sets of data in SQL Server quite a bit and did some performance testing on the Index on while inserting and the add it afterwards. I found that BY FAR it was much more efficient to create the index after all data was loaded. In our case it took 1 hour to load with the index added at the end, and 4 hours to add it with the index still on.
I think the key is to get the data moved as quick as possible, I am not sure if loading it in order really helps, do you have any stats on load time vs. index time? If you do, you could start to experiment a bit on that side of things.
Loading with the indexes dropped is better as a live index will generate several I/O's for every row in the database. 4 million rows is small enough that you would not expect to get a significant benefit from table partitioning.
You could get a performance win by using bcp to load the data into the staging area and running several tasks in parallel (SSIS will do this). Write a generic batch file wrapper for bcp that takes the file path (and table name if necessary) and invoke a series of jobs in half a dozen threads with 'Execute Process' tasks in SSIS. For 50 jobs it's probably not worth trying to write a data-driven load controller process. Wrap these tasks up in a sequence container so you don't have to maintain all of the dependencies explicitly.
You should definitely drop and re-create the indexes as this will greatly reduce the amount of I/O during the process.
If the 50 sources are being treated identically, try loading them into a common table or building a partitioned view over the staging tables.
Index at the end, yes. Also consider setting the log level setting to BULK LOGGED to minimize writes to the transaction log. Just remember to set it back to FULL after you've finished.
To the best of my knowledge, you are correct - it's much better to add the records all at once and then index once at the end.