Forcing Locality on Dask Dataframe Subsets - dask

I'm trying to distribute a large Dask Dataframe across multiple machines for (later) distributed computations on the dataframe. I'm using dask-distributed for this.
All the dask-distributed examples/docs I see are populating the initial data load from a network resource (hdfs, s3, etc) and does not appear to extend the DAG optimization to the load portion (seems to assume that a network load is a necessary evil and just eats the initial cost.) This is underscored on the answer to another question: Does Dask communicate with HDFS to optimize for data locality?
However, I can see cases where we would want this. For example, if we have a sharded database + dask workers co-located on nodes of this DB, we would want to force records from only the local shard to be populated into the local dask workers. From the documentation/examples, network cris-cross seems like a necessarily assumed cost. Is it possible to force parts of a single dataframe to be obtained from specific workers?
The alternative, which I've tried, is to try and force each worker to run a function (iteratively submitted to each worker) where the function loads only the data local to that machine/shard. This works, and I have a bunch of optimally local dataframes with the same column schema -- however -- now I don't have a single dataframe but n dataframes. Is it possible to merge/fuse dataframes across multiple machines so there is a single dataframe reference, but portions have affinity (within reason, as decided by the task DAG) to specific machines?

You can produce dask "collections" such as a dataframe from futures and delayed objects, which inter-operate nicely with each other.
For each partition, where you know which machine should load it, you can produce a future as follows:
f = c.submit(make_part_function, args, workers={'my.worker.ip'})
where c is the dask client and the address is the machine you'd want to see it happen on. You can also give allow_other_workers=True is this is a preference rather than a requirement.
To make a dataframe, from a list of such futures, you could do
df = dd.from_delayed([dask.delayed(f) for f in futures])
and ideally provide a meta=, giving a description of the expected dataframe. Now, further operations on a given partition will prefer to be scheduled on the same worker which already holds the data.

I am also interested in having the capability to restrict computation to a specific node (and data localized to that node). I have tried to implement the above with a simple script (see below) but looking at the resulting data frame, results the error (from dask/dataframe/utils.py::check_meta()):
ValueError: Metadata mismatch found in `from_delayed`.
Expected partition of type `DataFrame` but got `DataFrame`
Example:
from dask.distributed import Client
import dask.dataframe as dd
import dask
client = Client(address='<scheduler_ip>:8786')
client.restart()
filename_1 = 'http://samplecsvs.s3.amazonaws.com/Sacramentorealestatetransactions.csv'
filename_2 = 'http://samplecsvs.s3.amazonaws.com/SalesJan2009.csv'
future_1 = client.submit(dd.read_csv, filename_1, workers='w1')
future_2 = client.submit(dd.read_csv, filename_2, workers='w2')
client.has_what()
# Returns: {'tcp://<w1_ip>:41942': ('read_csv-c08b231bb22718946756cf46b2e0f5a1',),
# 'tcp://<w2_ip>:41942': ('read_csv-e27881faa0f641e3550a8d28f8d0e11d',)}
df = dd.from_delayed([dask.delayed(f) for f in [future_1, future_2]])
type(df)
# Returns: dask.dataframe.core.DataFrame
df.head()
# Returns:
# ValueError: Metadata mismatch found in `from_delayed`.
# Expected partition of type `DataFrame` but got `DataFrame`
Note The dask environment has a two worker nodes (aliased to w1 and w2) a scheduler node and the script is running on an external host.
dask==1.2.2, distributed==1.28.1

It is odd to call many dask dataframe functions in parallel. Perhaps you meant to call many Pandas read_csv calls in parallel instead?
# future_1 = client.submit(dd.read_csv, filename_1, workers='w1')
# future_2 = client.submit(dd.read_csv, filename_2, workers='w2')
future_1 = client.submit(pandas.read_csv, filename_1, workers='w1')
future_2 = client.submit(pandas.read_csv, filename_2, workers='w2')
See https://docs.dask.org/en/latest/delayed-best-practices.html#don-t-call-dask-delayed-on-other-dask-collections for more information

Related

dask.distributed not utilising the cluster

I'm not able to process this block using the distributed cluster.
import pandas as pd
from dask import dataframe as dd
import dask
df = pd.DataFrame({'reid_encod': [[1,2,3,4,5,6,7,8,9,10],[1,2,3,4,5,6,7,8,9,10],[1,2,3,4,5,6,7,8,9,10],[1,2,3,4,5,6,7,8,9,10],[1,2,3,4,5,6,7,8,9,10],[1,2,3,4,5,6,7,8,9,10]]})
dask_df = dd.from_pandas(df, npartitions=3)
save_val = []
def add(dask_df):
for _, outer_row in dask_df.iterrows():
for _, inner_row in dask_df.iterrows():
for base_encod in outer_row['reid_encod']:
for compare_encod in inner_row['reid_encod']:
val = base_encod + compare_encod
save_val.append(val)
return save_val
from dask.distributed import Client
client = Client(...)
dask_compute = dask.delayed(add)(dask_df)
dask_compute.compute()
Also I have few queries
Does dask.delayed use the available clusters to do the computation.
Can I paralleize the for loop iteratition of this pandas DF using delayed, and use multiple computers present in the cluster to do computations.
does dask.distributed work on pandas dataframe.
can we use dask.delayed in dask.distributed.
If the above programming approach is wrong, can you guide me whether to choose delayed or dask DF for the above scenario.
For the record, some answers, although I wish to note my earlier general points about this question
Does dask.delayed use the available clusters to do the computation.
If you have created a client to a distributed cluster, dask will use it for computation unless you specify otherwise.
Can I paralleize the for loop iteratition of this pandas DF using delayed, and use multiple computers present in the cluster to do computations.
Yes, you can in general use delayed with pandas dataframes for parallelism if you wish. However, your dataframe only has one row, so it is not obvious in this case how - it depends on what you really want to achieve.
does dask.distributed work on pandas dataframe.
Yes, you can do anything that python can do with distributed, since it is just python processes executing code. Whether it brings you the performance you are after is a separate question
can we use dask.delayed in dask.distributed.
Yes, distributed can execute anything that dask in general can, including delayed functions/objects
If the above programming approach is wrong, can you guide me whether to choose delayed or dask DF for the above scenario.
Not easily, it is not clear to me that this is a dataframe operation at all. It seems more like an array - but, again, I note that your function does not actually return anything useful at all.
In the tutorial: passing pandas dataframes to delayed ; same with dataframe API.
The main problem with your code is sketched in this section of the best practices: don't pass Dask collections to delayed functions. This means, you should use either the delayed API or the dataframe API. While you can convert dataframes<->delayed, simply passing like this is not recommended.
Furthermore,
you only have one row in your dataframe, so you only get one partition and no parallelism whatever. You can only slow things down like this.
this appears to be an everything-to-everything (N^2) operation, so if you had many rows (the normal case for Dask), it would presumably take extremely long, no matter how many cores you used
passing lists in a pandas row is not a great idea, perhaps you wanted to use an array?
the function doesn't return anything useful, so it's not at all clear what you are trying to achieve. Under the description of MVCE, you will see references to "expected outcome" and "what went wrong". To get more help, please be more precise.

Dask distributed perform computations without returning data

I have a dynamic Dask Kubernetes cluster.
I want to load 35 parquet files (about 1.2GB) from Gcloud storage into Dask Dataframe then process it with apply() and after saving the result to parquet file to Gcloud.
During loading files from Gcloud storage, a cluster memory usage is increasing to about 3-4GB. Then workers (each worker has 2GB of RAM) are terminated/restarted and some tasks getting lost,
so cluster starts computing the same things in a circle.
I removed apply() operation and leave only read_parquet() to test
if my custom code causes a trouble, but the problem was the same, even with just single read_parquet() operation. This is a code:
client = Client('<ip>:8786')
client.restart()
def command():
client = get_client()
df = dd.read_parquet('gcs://<bucket>/files/name_*.parquet', storage_options={'token':'cloud'}, engine='fastparquet')
df = df.compute()
x = client.submit(command)
x.result()
Note: I'm submitting a single command function to run all necessary commands to avoid problems with gcsfs authentication inside a cluster
After some investigation, I understood that problem could be in .compute() which returns all data to a process, but this process (my command function) is running on a worker. Because of that, a worker doesn't have enough RAM, crashes and lose all computed task which triggers tasks re-run.
My goal is:
to read from parquet files
perform some computations with apply()
and without even returning data from a cluster write it back to Gcloud storage in parquet format.
So, simply I want to keep data on a cluster and not return it back. Just compute and save data somewhere else.
After reading Dask distributed docs, I have found client.persist()/compute() and .scatter() methods. They look like what I need, but I don't really understand how to use them.
Could you, please, help me with client.persist() and client.compute() methods for my example
or suggest another way to do it? Thank you very much!
Dask version: 0.19.1
Dask distributed version: 1.23.1
Python version: 3.5.1
df = dd.read_parquet('gcs://<bucket>/files/name_*.parquet', storage_options={'token':'cloud'}, engine='fastparquet')
df = df.compute() # this triggers computations, but brings all of the data to one machine and creates a Pandas dataframe
df = df.persist() # this triggers computations, but keeps all of the data in multiple pandas dataframes spread across multiple machines

Streamz with Dask Distributed

Based on the streamz documentation, one could leverage a dask distributed cluster in the following way:
from distributed import Client
client = Client('tcp://localhost:8786') # Connect to scheduler that has distributed workers
from streamz import Stream
source = Stream()
(source.scatter() # scatter local elements to cluster, creating a DaskStream
.map(increment) # map a function remotely
.buffer(5) # allow five futures to stay on the cluster at any time
.gather() # bring results back to local process
.sink(write)) # call write locally
for x in range(10):
source.emit(x)
Conceptually, it isn't clear why we don't have to pass the dask distributed client in as a parameter to instantiate Stream(). More specifically, how does Stream() know what scheduler to attach to?
What would you do if you had two schedulers that have workers on unrelated nodes like:
from distributed import Client
client_1 = Client('tcp://1.2.3.4:8786')
client_2 = Client('tcp://10.20.30.40:8786')
How does one create two streams for client_1 and client_2, respectively?
The basic rule in Dask is, if there is a distributed client defined, use it for any Dask computations. If there is more than one distributed client, use the most recently created on that is still alive.
Streamz does not explicitly let you choose which client to use when you .scatter(), it uses dask.distributed.default_client() to pick one. You may wish to raise an issue with them to allow a client= keyword. The workflow doesn't even fit a context-based approach. For now, if you wanted to have simultaneous multiple streamz working with data on different Dask clusters, you would probably have to manipulate the state of dask.distributed.client._global_clients.

Batch results of intermediate dask computation

I have a large (10s of GB) CSV file that I want to load into dask, and for each row, perform some computation. I also want to write the results of the manipulated CSV into BigQuery, but it'd be better to batch network requests to BigQuery in groups of say, 10,000 rows each, so I don't incur network overhead per row.
I've been looking at dask delayed and see that you can create an arbitrary computation graph, but I'm not sure if this is the right approach: how do I collect and fire off intermediate computations based on some group size (or perhaps time elapsed). Can someone provide a simple example on that? Say for simplicity we have these functions:
def change_row(r):
# Takes 10ms
r = some_computation(r)
return r
def send_to_bigquery(rows):
# Ideally, in large-ish groups, say 10,000 rows at a time
make_network_request(rows)
# And here's how I'd use it
import dask.dataframe as dd
df = dd.read_csv('my_large_dataset.csv') # 20 GB
# run change_row(r) for each r in df
# run send_to_big_query(rows) for each appropriate size group based on change_row(r)
Thanks!
The easiest thing that you can do is provide a block size parameter to read_csv, which will get you approximately the right number of rows per block. You may need to measure some of your data or experiment to get this right.
The rest of your task will work the same way as any other "do this generic thing to blocks of data-frame": the `map_partitions' method (docs).
def alter_and_send(df):
rows = [change_row(r) for r in df.iterrows()]
send_to_big_query(rows)
return df
df.map_partitions(alter_and_send)
Basically, you are running the function on each piece of the logical dask dataframe, which are real pandas dataframes.
You may actually want map, apply or other dataframe methods in the function.
This is one way to do it - you don't really need the "output" of the map, and you could have used to_delayed() instead.

Parallelization on cluster dask

I'm looking for the best way to parallelize on a cluster the following problem. I have several files
folder/file001.csv
folder/file002.csv
:
folder/file100.csv
They are disjoints with respect to the key I want to use to groupby, that is if a set of keys is in file1.csv any of these keys has an item in any other files.
In one side I can just run
df = dd.read_csv("folder/*")
df.groupby("key").apply(f, meta=meta).compute(scheduler='processes')
But I'm wondering if there is a better/smarter way to do so in a sort of
delayed-groupby way.
Every filexxx.csv fits in memory on a node. Given that every node has n cores it will be ideal use all of them. For every single file I can use this hacky way
import numpy as np
import multiprocessing as mp
cores = mp.cpu_count() #Number of CPU cores on your system
partitions = cores #Define as many partitions as you want
def parallelize(data, func):
data_split = np.array_split(data, partitions)
pool = mp.Pool(cores)
data = pd.concat(pool.map(func, data_split))
pool.close()
pool.join()
return data
data = parallelize(data, f);
And, again, I'm not sure if there is an efficent dask way to do so.
you could use a Client (will run in multi process by default) and read your data with a certain blocksize. you can get the amount of workers (and number of cores per worker) with the ncores method and then calculate the optimal blocksize.
however according to the documantaion blocksize is by default "computed based on available physical memory and the number of cores."
so i think the best way to do it is a simple:
from distributed import Client
# if you run on a single machine just do: client = Client()
client = Client('cluster_scheduler_path')
ddf = dd.read_csv("folder/*")
EDIT: after that use map_partitions and do the gorupby for each partition:
# Note ddf is a dask dataframe and df is a pandas dataframe
new_ddf = ddf.map_partitions(lambda df: df.groupby("key").apply(f), meta=meta)
don't use compute because it will result in a single pandas.dataframe, instead use a dask output method to keep the entire process parallel and larger then ram compatible.

Resources