RE: Hash Partitioning and Dataframes
Just trying to make sure that something I know in advance (the joins will always have an equality test on one specific field) is used to optimize the partitioning so the joins only use local data. Thanks for the info. Ron From: Michael Armbrust [mailto:mich...@databricks.com] Sent: Friday, May 08, 2015 3:15 PM To: Daniel, Ronald (ELS-SDG) Cc: user@spark.apache.org Subject: Re: Hash Partitioning and Dataframes What are you trying to accomplish? Internally Spark SQL will add Exchange operators to make sure that data is partitioned correctly for joins and aggregations. If you are going to do other RDD operations on the result of dataframe operations and you need to manually control the partitioning, call df.rdd and partition as you normally would. On Fri, May 8, 2015 at 2:47 PM, Daniel, Ronald (ELS-SDG) r.dan...@elsevier.commailto:r.dan...@elsevier.com wrote: Hi, How can I ensure that a batch of DataFrames I make are all partitioned based on the value of one column common to them all? For RDDs I would partitionBy a HashPartitioner, but I don't see that in the DataFrame API. If I partition the RDDs that way, then do a toDF(), will the partitioning be preserved? Thanks, Ron
RE: How to do spares vector product in Spark?
Any convenient tool to do this [sparse vector product] in Spark? Unfortunately, it seems that there are very few operations defined for sparse vectors. I needed to add some, and ended up converting them to (dense) numpy vectors and doing the addition on those. Best regards, Ron From: Xi Shen [mailto:davidshe...@gmail.com] Sent: Friday, March 13, 2015 1:50 AM To: user@spark.apache.org Subject: How to do spares vector product in Spark? Hi, I have two RDD[Vector], both Vector are spares and of the form: (id, value) id indicates the position of the value in the vector space. I want to apply dot product on two of such RDD[Vector] and get a scale value. The none exist values are treated as zero. Any convenient tool to do this in Spark? Thanks, David
sparse vector operations in Python
Hi, Sorry to ask this, but how do I compute the sum of 2 (or more) mllib SparseVectors in Python? Thanks, Ron - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
partions, SQL tables, and Parquet I/O
Short story: I want to write some parquet files so they are pre-partitioned by the same key. Then, when I read them back in, joining the two tables on that key should be about as fast as things can be done. Can I do that, and if so, how? I don't see how to control the partitioning of a SQL table, as opposed to PairRDDs. Thanks, Ron
RE: Using TF-IDF from MLlib
Thanks for the info Andy. A big help. One thing - I think you can figure out which document is responsible for which vector without checking in more code. Start with a PairRDD of [doc_id, doc_string] for each document and split that into one RDD for each column. The values in the doc_string RDD get split and turned into a Seq and fed to TFIDF. You can take the resulting RDD[Vector]s and zip them with the doc_id RDD. Presto! Best regards, Ron
Using TF-IDF from MLlib
Hi all, I want to try the TF-IDF functionality in MLlib. I can feed it words and generate the tf and idf RDD[Vector]s, using the code below. But how do I get this back to words and their counts and tf-idf values for presentation? val sentsTmp = sqlContext.sql(SELECT text FROM sentenceTable) val documents: RDD[Seq[String]] = sentsTmp.map(_.toString.split( ).toSeq) val hashingTF = new HashingTF() val tf: RDD[Vector] = hashingTF.transform(documents) tf.cache() val idf = new IDF().fit(tf) val tfidf: RDD[Vector] = idf.transform(tf) It looks like I can get the indices of the terms using something like J = wordListRDD.map(w = hashingTF.indexOf(w)) where wordList is an RDD holding the distinct words from the sequence of words used to come up with tf. But how do I do the equivalent of Counts = J.map(j = tf.counts(j)) ? Thanks, Ron
RE: filtering a SchemaRDD
Indeed it did. Thanks! Ron From: Michael Armbrust [mailto:mich...@databricks.com] Sent: Friday, November 14, 2014 9:53 PM To: Daniel, Ronald (ELS-SDG) Cc: user@spark.apache.org Subject: Re: filtering a SchemaRDD If I use row[6] instead of row[text] I get what I am looking for. However, finding the right numeric index could be a pain. Can I access the fields in a Row of a SchemaRDD by name, so that I can map, filter, etc. without a trial and error process of finding the right int for the fieldname? row.text should work. More examples here: http://spark.apache.org/docs/1.1.0/sql-programming-guide.html#tab_python_2 Michael
filtering a SchemaRDD
Hi all, I have a SchemaRDD that Is loaded from a file. Each Row contains 7 fields, one of which holds the text for a sentence from a document. # Load sentence data table sentenceRDD = sqlContext.parquetFile('s3n://some/path/thing') sentenceRDD.take(3) Out[20]: [Row(annotID=118, annotSet=u'ge', annotType=u'sentence', endOffset=20194, pii=u'0094576587900440', startOffset=20062, text=u'Paper IAF-86-85 presented at the 37th Congress of the International Astronautical Federation, Innsbruck, Austria, 4-11 October 1986.'), Row(annotID=163, annotSet=u'ge', annotType=u'sentence', endOffset=20249, pii=u'0094576587900440', startOffset=20194, text=uThe landsat sensors: Eosat's plans for landsats 6 and 7), Row(annotID=190, annotSet=u'ge', annotType=u'sentence', endOffset=20342, pii=u'0094576587900440', startOffset=20334, text=u'Abstract')] I have this registered as a table and can query it with SQL select statments. I would also like to filter the RDD using text operations like regexps that have greated capabilities than SQL's LIKE operator. However, the code below does not work. Instead I get a runtime error. openProbsRDD = sentenceRDD.filter(lambda row: remains unknown in row[text] ) openProbsRDD.take(5) ... TypeError: tuple indices must be integers, not str ... If I use row[6] instead of row[text] I get what I am looking for. However, finding the right numeric index could be a pain. Can I access the fields in a Row of a SchemaRDD by name, so that I can map, filter, etc. without a trial and error process of finding the right int for the fieldname? Thanks, Ron Daniel
Accessing neighboring elements in an RDD
Hi all, Assume I have read the lines of a text file into an RDD: textFile = sc.textFile(SomeArticle.txt) Also assume that the sentence breaks in SomeArticle.txt were done by machine and have some errors, such as the break at Fig. in the sample text below. Index Text N...as shown in Fig. N+1 1. N+2 The figure shows... What I want is an RDD with: N ... as shown in Fig. 1. N+1 The figure shows... Is there some way a filter() can look at neighboring elements in an RDD? That way I could look, in parallel, at neighboring elements in an RDD and come up with a new RDD that may have a different number of elements. Or do I just have to sequentially iterate through the RDD? Thanks, Ron
RE: Accessing neighboring elements in an RDD
Thanks for the pointer to that thread. Looks like there is some demand for this capability, but not a lot yet. Also doesn't look like there is an easy answer right now. Thanks, Ron From: Victor Tso-Guillen [mailto:v...@paxata.com] Sent: Wednesday, September 03, 2014 10:40 AM To: Daniel, Ronald (ELS-SDG) Cc: user@spark.apache.org Subject: Re: Accessing neighboring elements in an RDD Interestingly, there was an almost identical question posed on Aug 22 by cjwang. Here's the link to the archive: http://apache-spark-user-list.1001560.n3.nabble.com/Finding-previous-and-next-element-in-a-sorted-RDD-td12621.html#a12664 On Wed, Sep 3, 2014 at 10:33 AM, Daniel, Ronald (ELS-SDG) r.dan...@elsevier.commailto:r.dan...@elsevier.com wrote: Hi all, Assume I have read the lines of a text file into an RDD: textFile = sc.textFile(SomeArticle.txt) Also assume that the sentence breaks in SomeArticle.txt were done by machine and have some errors, such as the break at Fig. in the sample text below. Index Text N...as shown in Fig. N+1 1. N+2 The figure shows... What I want is an RDD with: N ... as shown in Fig. 1. N+1 The figure shows... Is there some way a filter() can look at neighboring elements in an RDD? That way I could look, in parallel, at neighboring elements in an RDD and come up with a new RDD that may have a different number of elements. Or do I just have to sequentially iterate through the RDD? Thanks, Ron
RE: Accessing neighboring elements in an RDD
Thanks Xiangrui, that looks very helpful. Best regards, Ron -Original Message- From: Xiangrui Meng [mailto:men...@gmail.com] Sent: Wednesday, September 03, 2014 1:19 PM To: Daniel, Ronald (ELS-SDG) Cc: Victor Tso-Guillen; user@spark.apache.org Subject: Re: Accessing neighboring elements in an RDD There is a sliding method implemented in MLlib (https://github.com/apache/spark/blob/master/mllib/src/main/scala/org/a pache/spark/mllib/rdd/SlidingRDD.scala), which is used in computing Area Under Curve: https://github.com/apache/spark/blob/master/mllib/src/main/scala/org/a pache/spark/mllib/evaluation/AreaUnderCurve.scala#L45 With it, you can process neighbor lines by rdd.sliding(3).map { case Seq(l0, l1, l2) = ... } -Xiangrui On Wed, Sep 3, 2014 at 11:30 AM, Daniel, Ronald (ELS-SDG) r.dan...@elsevier.com wrote: Thanks for the pointer to that thread. Looks like there is some demand for this capability, but not a lot yet. Also doesn't look like there is an easy answer right now. Thanks, Ron From: Victor Tso-Guillen [mailto:v...@paxata.com] Sent: Wednesday, September 03, 2014 10:40 AM To: Daniel, Ronald (ELS-SDG) Cc: user@spark.apache.org Subject: Re: Accessing neighboring elements in an RDD Interestingly, there was an almost identical question posed on Aug 22 by cjwang. Here's the link to the archive: http://apache-spark-user-list.1001560.n3.nabble.com/Finding-previous-a nd-next-element-in-a-sorted-RDD-td12621.html#a12664 On Wed, Sep 3, 2014 at 10:33 AM, Daniel, Ronald (ELS-SDG) r.dan...@elsevier.com wrote: Hi all, Assume I have read the lines of a text file into an RDD: textFile = sc.textFile(SomeArticle.txt) Also assume that the sentence breaks in SomeArticle.txt were done by machine and have some errors, such as the break at Fig. in the sample text below. Index Text N...as shown in Fig. N+1 1. N+2 The figure shows... What I want is an RDD with: N ... as shown in Fig. 1. N+1 The figure shows... Is there some way a filter() can look at neighboring elements in an RDD? That way I could look, in parallel, at neighboring elements in an RDD and come up with a new RDD that may have a different number of elements. Or do I just have to sequentially iterate through the RDD? Thanks, Ron
RE: Regarding tooling/performance vs RedShift
Just to point out that the benchmark you point to has Redshift running on HDD machines instead of SSD, and it is still faster than Shark in all but one case. Like Gary, I'm also interested in replacing something we have on Redshift with Spark SQL, as it will give me much greater capability to process things. I'm willing to sacrifice some performance for the greater capability. But it would be nice to see the benchmark updated with Spark SQL, and with a more competitive configuration of Redshift. Best regards, and keep up the great work! Ron From: Nicholas Chammas [mailto:nicholas.cham...@gmail.com] Sent: Wednesday, August 06, 2014 9:30 AM To: Gary Malouf Cc: user Subject: Re: Regarding tooling/performance vs RedShift 1) We get tooling out of the box from RedShift (specifically, stable JDBC access) - Spark we often are waiting for devops to get the right combo of tools working or for libraries to support sequence files. The arguments about JDBC access and simpler setup definitely make sense. My first non-trivial Spark application was actually an ETL process that sliced and diced JSON + tabular data and then loaded it into Redshift. From there on you got all the benefits of your average C-store database, plus the added benefit of Amazon managing many annoying setup and admin details for your Redshift cluster. One area I'm looking forward to seeing Spark SQL excel at is offering fast JDBC access to raw data--i.e. directly against S3 / HDFS; no ETL required. For easy and flexible data exploration, I don't think you can beat that with a C-store that you have to ETL stuff into. 2) There is a belief that for many of our queries (assumed to often be joins) a columnar database will perform orders of magnitude better. This is definitely a it depends statement, but there is a detailed benchmark herehttps://amplab.cs.berkeley.edu/benchmark/ comparing Shark, Redshift, and other systems. Have you seen it? Redshift does very well, but Shark is on par or better than it in most of the tests. Of course, going forward we'll want to see Spark SQL match this kind of performance, and that remains to be seen. Nick On Wed, Aug 6, 2014 at 12:06 PM, Gary Malouf malouf.g...@gmail.commailto:malouf.g...@gmail.com wrote: My company is leaning towards moving much of their analytics work from our own Spark/Mesos/HDFS/Cassandra set up to RedShift. To date, I have been the internal advocate for using Spark for analytics, but a number of good points have been brought up to me. The reasons being pushed are: - RedShift exposes a jdbc interface out of the box (no devops work there) and data looks and feels like it is in a normal sql database. They want this out of the box from Spark, no trying to figure out which version matches this version of Hive/Shark/SparkSQL etc. Yes, the next release theoretically supports this but there have been release issues our team has battled to date that erode the trust. - Complaints around challenges we have faced running a spark shell locally against a cluster in EC2. It is partly a devops issue of deploying the correct configurations to local machines, being able to kick a user off hogging RAM, etc. - I want to be able to run queries from my python shell against your sequence file data, roll it up and in the same shell leverage python graph tools. - I'm not very familiar with the Python setup, but I believe by being able to run locally AND somehow add custom libraries to be accessed from PySpark this could be done. - Joins will perform much better (in RedShift) because it says it sorts it's keys. We cannot pre-compute all joins away. Basically, their argument is two-fold: 1) We get tooling out of the box from RedShift (specifically, stable JDBC access) - Spark we often are waiting for devops to get the right combo of tools working or for libraries to support sequence files. 2) There is a belief that for many of our queries (assumed to often be joins) a columnar database will perform orders of magnitude better. Anyway, a test is being setup to compare the two on the performance side but from a tools perspective it's hard to counter the issues that are brought up.
RE: Regarding tooling/performance vs RedShift
Well yes, MLlib-like routines or pretty much anything else could be run on the derived results, but you have to unload the results from Redshift and then load them into some other tool. So it's nicer to leave them in memory and operate on them there. Major architectural advantage to Spark. Ron From: Gary Malouf [mailto:malouf.g...@gmail.com] Sent: Wednesday, August 06, 2014 1:17 PM To: Nicholas Chammas Cc: Daniel, Ronald (ELS-SDG); user@spark.apache.org Subject: Re: Regarding tooling/performance vs RedShift Also, regarding something like redshift not having MLlib built in, much of that could be done on the derived results. On Aug 6, 2014 4:07 PM, Nicholas Chammas nicholas.cham...@gmail.commailto:nicholas.cham...@gmail.com wrote: On Wed, Aug 6, 2014 at 3:41 PM, Daniel, Ronald (ELS-SDG)r.dan...@elsevier.commailto:r.dan...@elsevier.com wrote: Mostly I was just objecting to Redshift does very well, but Shark is on par or better than it in most of the tests when that was not how I read the results, and Redshift was on HDDs. My bad. You are correct; the only test Shark (mem) does better on is test #1 Scan Query. And indeed, it would be good to see an updated benchmark with Redshift running on SSDs. Nick
Column width limits?
Assume I want to make a PairRDD whose keys are S3 URLs and whose values are Strings holding the contents of those (UTF-8) files, but NOT split into lines. Are there length limits on those files/Strings? 1 MB? 16 MB? 4 GB? 1 TB? Similarly, can such a thing be registered as a table so that I can use substr() to pick out pieces of the string? Thanks, Ron