I created a database containing a total of 3 tables for a specific purpose. The total size of all tables is about 850 MB - very lean... out of which one single table contains about 800 MB (including index) of data and 5 million records (daily addition of about 6000 records).
The system is PG-Windows with 8 GB RAM Windows 7 laptop with SSD.
I allocated 2048MB as shared_buffers, 256MB as temp_buffers and 128MB as work_mem.
I execute a single query multiple times against the single table - hoping that the table stays in RAM (hence the above parameters).
But, although I see a spike in memory usage during execution (by about 200 MB), I do not see memory consumption remaining at at least 500 MB (for the data to stay in memory). All postgres exe running show 2-6 MB size in task manager. Hence, I suspect the LRU does not keep the data in memory.
Average query execution time is about 2 seconds (very simple single table query)... but I need to get it down to about 10-20 ms or even lesser if possible, purely because there are just too many times, the same is going to be executed and can be achieved only by keeping stuff in memory.
Any advice?
Regards,
Kapil
You should not expect postgres processes to show large memory use, even if the whole database is cached in RAM.
That is because PostgreSQL relies on buffered reads from the operating system buffer cache. In simplified terms, when PostgreSQL does a read(), the OS looks to see whether the requested blocks are cached in the "free" RAM that it uses for disk cache. If the block is in cache, the OS returns it almost instantly. If the block is not in cache the OS reads it from disk, adds it to the disk cache, and returns the block. Subsequent reads will fetch it from the cache unless it's displaced from the cache by other blocks.
That means that if you have enough free memory to fit the whole database in "free" operating system memory, you won't tend to hit the disk for reads.
Depending on the OS, behaviour for disk writes may differ. Linux will write-back cache "dirty" buffers, and will still return blocks from cache even if they've been written to. It'll write these back to the disk lazily unless forced to write them immediately by an fsync() as Pg uses at COMMIT time. When it does that it marks the cached blocks clean, but doesn't flush them. I don't know how Windows behaves here.
The point is that PostgreSQL can be running entirely out of RAM with a 1GB database, even though no PostgreSQL process seems to be using much RAM. Having shared_buffers too high just leads to double-caching and can reduce the amount of RAM available for the OS to cache blocks.
It isn't easy to see exactly what's cached in RAM because Pg relies on the OS cache. That's why I referred you to pg_fincore.
If you're on Windows and this won't work, you really just have to rely on observing disk activity. Does performance monitor show lots of uncached disk reads? Does operating system memory monitoring show lots of memory used for disk cache in the OS?
Make sure that effective_cache_size correctly reflects the RAM used for disk cache. It will help PostgreSQL choose appropriate query plans.
You are making the assumption, without apparent evidence, that the query performance you are experiencing is explained by disk read delays, and that it can be improved by in-memory caching. This may not be the case at all. You need to look at explain analyze output and system performance metrics to see what's going on.
Related
I have been considering the idea of moving to a RAMdisk for a while. I know its risks, but just wanted to do a little benchmark. I just had two questions: (a) when reading the query plan, will it still differentiate between disk and buffers hits? If so, should I assume that both are equally expensive or should I assume that there is a difference between them?
(b) a RAM disk is not persistent, but if I want to export some results to persistent storage, are there some precautions I would need to take? Is it the same as usual e.g. COPY command?
I do not recommend using RAM disks in PostgreSQL for persistent storage. With careful tuning, you can get PostgreSQL not to use more disk I/O than what is required to make your data persistent.
I recommend doing this:
Have more RAM in your machine than the size of the database.
Define shared_buffers big enough to contain the database (on Linux, define memory hugepages to contain them).
Increase checkpoint_timeout and max_wal_size to get fewer checkpoints.
Set synchronous_commit = off to keep PostgreSQL from syncing WAL to disk on every commit.
If you are happy to lose all your data in the case of a crash, define your tables UNLOGGED. The data will survive a normal shutdown.
Anyway, to answer your questions:
(a) You should set seq_page_cost and random_page_cost way lower to tell PostgreSQL how fast your storage is.
(b) You could run backups with either pg_dump or pg_basebackup, they don't care what kind of storage you have got.
when reading the query plan, will it still differentiate between disk and buffers hits?
It never distinguished between them in the first place. It distinguishes between "hit" and "read", but the "read" can't tell which are truly from disk and which are from OS/FS cache.
PostgreSQL has no idea you are running on a RAM disk, so will continue to report those as it always has.
If so, should I assume that both are equally expensive or should I assume that there is a difference between them?
This is a question that should be answered through your benchmarking. On some systems, memory can be read-ahead from main memory into the faster caches, making sequential reads still faster than random reads. If you care, you will have to benchmark it on your own system.
Reading data from RAM into shared_buffers is still surprisingly expensive due to things like lock management. So as a rough starting point, maybe seq_page_cost=0.1 and random_page_cost=0.15.
a RAM disk is not persistent, but if I want to export some results to persistent storage, are there some precautions I would need to take?
The risk would be that your system crashes before the export has finished. But what precaution can you take against that?
Say I have a single collection in mongodb with only one index, and I require the index for the entire life cycle of the application using that mongo collection.
I would like to know about the behaviour of mongodb.
In this case once the index is loaded into memory, will mongodb keep it in the ram?
Thanks
The first thing MongoDB will knock out of RAM will be the LRU (least recently used) piece of data. So if you only have one index, chances are it will continue to be used pretty regularly and it should stay in memory.
Source
Unfortunately you cannot currently pin a collection or index in memory. MongoDB uses memory mapped files to load collections and indexes into memory. As your activities touch various pieces of your database thru queries, updates, insertions and deletions, that data will get loaded into memory. This is referred to as the working set. If the total memory required to load the working set is less than available memory, no problem.
If not, MongoDB is going to use an LRU algorithm to pick what to unload from memory. This is why it's so important to understand the concept of the working set and how it relates to your available memory.
This writeup from the documentation should be helpful:
How do I calculate how much RAM I need for my application?
The amount of RAM you need depends on several factors, including but
not limited to:
The relationship between database storage and working set.
The operating system’s cache strategy for LRU (Least Recently Used)
The impact of journaling
The number or rate of page faults and other MMS gauges to detect when you need more RAM
Each database connection thread will need up to 1 MB of RAM. MongoDB
defers to the operating system when loading data into memory from
disk. It simply memory maps all its data files and relies on the
operating system to cache data. The OS typically evicts the
least-recently-used data from RAM when it runs low on memory. For
example if clients access indexes more frequently than documents, then
indexes will more likely stay in RAM, but it depends on your
particular usage.
To calculate how much RAM you need, you must calculate your working
set size, or the portion of your data that clients use most often.
This depends on your access patterns, what indexes you have, and the
size of your documents. Because MongoDB uses a thread per connection
model, each database connection also will need up to 1MB of RAM,
whether active or idle.
If page faults are infrequent, your working set fits in RAM. If fault
rates rise higher than that, you risk performance degradation. This is
less critical with SSD drives than with spinning disks.
http://docs.mongodb.org/manual/faq/diagnostics/
You can use the serverStatus command to get an estimate of your current working set:
db.runCommand( { serverStatus: 1, workingSet: 1 } )
After starting Mongo via mongod, I ran a Mongo query that took 300 seconds. Calling db.serverStatus() on my "admin" db showed Mongo having resident memory of 1 GB. The docs explain that "resident" memory is the amount of physical disk/RAM that Mongo uses.
Then, I re-ran the same query, but it took 8 seconds. Looking at the resident memory this time, I saw 5 GB.
The large increase in RAM, I believe, helps to explain why the query time shrank from 300 to 8 seconds, but why did the resident memory jump so quickly?
Is there some type of "warming" step recommended to prepare Mongo so as to avoid 300 second queries?
There reason behind that MongoDB uses mmap functionality of the operating system. This means, at least on Linux systems That the memory handling of the mongodb is based on some functionality of the operating system called memory mapped files.
The memory in Linux systems is addressed in several levels basically any program will see an address space on 32-bit systems of 2GB over all, on 64-bit systems 128TB. This is a virtual address space which means on 32/64bit that amount of memory can be addressed with 4kb memory pages(page is the individually handled part of the memory). That is why if you start mongoDB on a 32 bit system it will rise a warning that the database on such system can handle only 2GB of data. Obviously this virtual address space is bigger than the amount of physical memory so there is a mapping between these virtual addresses and the physical ones. Some of the virtual addresses reside in really physical memory so they are in the real memory,but the algorithm which ensures this is on the side of the kernel. Programs running on Linux systems can deal only with virtual addresses, if one tries to access a virtual memory address which is not in physical memory, a pagefault occurs (you can track this on the serverStatus commands extra info field). (You can find short explanation of this here)
Accessing memory in case if the virtual address reside in physical memory is as fast as the memory, accessing a virtual address which has no physical currently means a paging from disk to memory and read the memory so as fast as the disks random read. (This makes the different in your case)
There is a command in mongoDB which with you can enforce the caching of a collection or an index this command is the touch
If you use this command to load the data into memory before the first query you will get the results in 8sec at first try. Unfortunately you cannot really force the OS to keep always this in memory, so if you have others things using up the memory OS will page out this data in some time.
IF you have enough physical memory mongoDB will keep everything the data and indexes in memory. This not always needed. There is a portion of data which need to be in memory to avoid extensive amount of pagefaults this is the workingset. You can check the size of the working set with the db.runCommand( { serverStatus: 1, workingSet: 1 } ) command.
You cannot handle the paging while it is OS level, but if you have enough memory usually the kernel likes to keep as much stuff cached as it can. If the workingset fits in memory you are more or less ok. If some documents really rarely accessed and there is not enough memory to keep everything there they will be paged out anyway.
When you run a query several things can happen. An index can cover which means no documents will be touched at all, if your query is selective in some notion only a part of the index will be touched. unfortunately it is really hard to define memory is sufficient and the only thing what you can do is to monitor (the workingset metric is an estimation). The symptom of running out of memory can be identified check this presentation. And use MMS.
I'm using MongoDB on a 32 bit production system, which sucks but it's out of my control right now. The challenge is to keep the memory usage under ~2.5GB since going over this will cause 32 bit systems to crash.
According to the mongoDB team, the best way to track the memory usage is to use your operating system's process tracking system (i.e. ps or htop on Unix systems; Process Explorer on Windows.) for virtual memory size.
The DB mainly consists of one table which is continually cycling data, i.e. receiving data at regular intervals from sensors, and every day a cron job wipes all data from before the last 3 days. Over a period of time, the memory usage slowly increases. I took some notes over time using db.serverStats(), db.lectura.totalSize() and ps, shown in the chart below. Note that the size of the table in question has reduced in the last month but the memory usage increased nonetheless.
Now, there is some scope for adjustment in how many days of data I store. Today I deleted basically half of the data, and then restarted mongodb, and yet the mem virtual / mem mapped and most importantly memory usage according to ps have hardly changed! Why do these not reduce when I wipe data (and restart)? I read some other questions where people said that mongo isn't really using all the memory that it might appear to be using, and that you can't clear the cache or limit memory use. But then how can I ensure I stay under the 2.5GB limit?
Unless there is a way to stem this dataset-size-irrespective gradual increase in memory usage, it seems to me that the 32-bit version of Mongo is unuseable. Note: I don't mind losing a bit of performance if it solves the problem.
To answer regarding why the mapped and virtual memory usage does not decrease with the deletes, the mapped number is actually what you get when you mmap() the entire set of data files. This does not shrink when you delete records, because although the space is freed up inside the data files, they are not themselves reduced in size - the files are just more empty afterwards.
Virtual will include journal files, and connections, and other non-data related memory usage also, but the same principle applies there. This, and more, is described here:
http://www.mongodb.org/display/DOCS/Checking+Server+Memory+Usage
So, the 2GB storage size limitation on 32-bit will actually apply to the data files whether or not there is data in them. To reclaim deleted space, you will have to run a repair. This is a blocking operation and will require the database to be offline/unavailable while it was run. It will also need up to 2x the original size in terms of free disk space to be able to run the repair, since it essentially represents writing out the files again from scratch.
This limitation, and the problems it causes, is why the 32-bit version should not be run in production, it is just not suitable. I would recommend getting onto a 64-bit version as soon as possible.
By the way, neither of these figures (mapped or virtual) actually represents your resident memory usage, which is what you really want to look at. The best way to do this over time is via MMS, which is the free monitoring service provided by 10gen - it will graph virtual, mapped and resident memory for you over time as well as plenty of other stats.
If you want an immediate view, run mongostat and check out the corresponding memory columns (res, mapped, virtual).
In general, when using 64-bit builds with essentially unlimited storage, the data will usually greatly exceed the available memory. Therefore, mongod will use all of the available memory it can in terms of resident memory (which is why you should always have swap configured to the OOM Killer does not come into play).
Once that is used, the OS does not stop allocating memory, it will just have the oldest items paged out to make room for the new data (LRU). In other words, the recycling of memory will be done for you, and the resident memory level will remain fairly constant.
Your options for stretching 32-bit are limited, but you can try some things. The thing that you run out of is address space, and the increases in the sizes of additional database files mean that you would like to avoid crossing over the boundary from "n" files to "n+1". It may be worth structuring your data into more or fewer databases so that you can get the maximum amount of actual data into memory and as little as possible "dead space".
For example, if your database named "mydatabase" consists of the files mydatabase.ns (the namespace file) at 16 MB, mydatabase.0 at 64 MB, mydatabase.1 at 128 MB and mydatabase.2 at 256 MB, then the next file created for this database will be mydatabase.3 at 512 MB. If instead of adding to mydatabase you instead created an additional database "mynewdatabase" it would start life with mynewdatabase.ns at 16 MB and mynewdatabase.0 at 64 MB ... quite a bit smaller than the 512 MB that adding to the original database would be. In fact, you could create 4 new databases for less space than would be consumed by adding a new file to the original database, and because the files are smaller they would be easier to fit into contiguous blocks of memory.
It is a well-known message that 32-bit should not be used for production.
Use 64-bit systems.
Point.
I am considering developing an application with a Cassandra backend. I am hoping that I will be able to run each cassandra node on commodity hardware with the following specs:
Quad Core 2GHz i7 CPU
2x 750GB disk drives
16 GB installed RAM
Now, I have been reading online that the available disk-space for Cassandra should be double the amount that is stored on the disks, which would mean that each node (set up in a RAID-1 configuration) would be able to store 375 GB of data, which is acceptable.
My question is this if 16GB RAM is enough to efficiently serve 375 GB of data per node. The data in the application developed will also be fairly time-dependant, such that recent data will be the data most read from the database. In fact, most of the data will be deleted after about 6 months.
Also, would I assign Cassandra a Heap (-Xmx) close to 16 GB, or does Cassandra utilize off-heap memory ?
You should not set the Cassandra heap to more than 8GB; bigger than that, and garbage collection will kill you with large pauses. Cassandra will use the buffer cache (like other applications) so the remaining memory isn't wasted.
16GB of RAM will be enough to serve the data if your hot set will all fit in RAM, or if serving rate can be served off disk. Disks can do about 100 random IO/s, so with your setup if you need more than 200 reads / second you will need to make sure the data is in cache. Cassandra exports good cache statistics (cassandra-cli show keyspaces) so you should easily be able to tell how effective your cache is being.
Do bear in mind, with only two disks in RAID-1, you will not have a dedicated commit log. This could hamper write performance quite badly. You may want to consider turning off the commit log if it does affect performance, and forgo durable writes.
Although it is probably wise not to use a really huge heap with Cassandra, at my company we have used 10GB to 12GB heaps without any issues so far. Our servers typically have at least 48 GB of memory (RAM is cheap -- so why not :-)) and so we may try expanding the heap a bit more and see what happens.