Sync job is very slow

  • Is this your first time deploying Airbyte?: Yes
  • OS Version / Instance: Linux (ubuntu 20.04) (Azure VM)
  • Memory / Disk: 8Gb / 32 Gb
  • Deployment: Docker
  • Airbyte Version: 0.40.26
  • Source name/version: S3 version 0.1.27
  • Destination name/version: Databricks version 0.3.1
  • Step: sync
  • Description:

Hi,

I am currently trying to set up a connection between AWS S3 and Databricks on Azure. I got the sync to work but it is very slow and reads much more data than there is to begin with.

  1. The sync takes approx. 140s for ca. 300MB and 200 000 rows which means a transfer rate of ~2.15MB/s and ~1500rows/s. I also tried a larger file with 2GB which took approx. 1h 15min which is a similar rate. Is this in the order of the expected transfer speed? It seems very slow to me. I also tried to speed it up by increasing the block size but I didn’t notice any change in speed when trying different block sizes. The reported speeds used a block size of 1 000 000. Is there anything else I could try to speed up the transfer?

  2. I am trying to transfer 200 000 entries from some demo data. The CSV-file on S3 is 71.4MB but according to the logs airbyte reads 297MB. As far as I understand airbyte this is because the data is internally converted to JSON which could result in a larger file size. But I did not expect it to be more than 4x as large. Is this how it is designed?

I attached the log of the mentioned sync run but removed the last part of the sync job because it fails afterwards but I already submitted another issue for this and I don’t think it is relevant here. (A complete version of the log is attached to the issue: S3 -> Databricks: Datatypes missing in SQL CREATE-statement · Issue #21193 · airbytehq/airbyte · GitHub)
airbyte_log_s3_to_databricks (2).txt (71.6 KB)

Hello there! You are receiving this message because none of your fellow community members has stepped in to respond to your topic post. (If you are a community member and you are reading this response, feel free to jump in if you have the answer!) As a result, the Community Assistance Team has been made aware of this topic and will be investigating and responding as quickly as possible.
Some important considerations that will help your to get your issue solved faster:

  • It is best to use our topic creation template; if you haven’t yet, we recommend posting a followup with the requested information. With that information the team will be able to more quickly search for similar issues with connectors and the platform and troubleshoot more quickly your specific question or problem.
  • Make sure to upload the complete log file; a common investigation roadblock is that sometimes the error for the issue happens well before the problem is surfaced to the user, and so having the tail of the log is less useful than having the whole log to scan through.
  • Be as descriptive and specific as possible; when investigating it is extremely valuable to know what steps were taken to encounter the issue, what version of connector / platform / Java / Python / docker / k8s was used, etc. The more context supplied, the quicker the investigation can start on your topic and the faster we can drive towards an answer.
  • We in the Community Assistance Team are glad you’ve made yourself part of our community, and we’ll do our best to answer your questions and resolve the problems as quickly as possible. Expect to hear from a specific team member as soon as possible.

Thank you for your time and attention.
Best,
The Community Assistance Team

Hello MoritzLampert, it’s been a while without an update from us. Are you still having problems or did you find a solution?

Hi,
no, I do not have any answers to my problems yet. One thing I did find was that the problem is not caused by my setup because the transfer rates when using Airbyte Cloud are comparable:

S3 → Azure Blob: 2,54 MB/s (Almost 220GB per 24h)
S3 → Databricks: 2,27 MB/s (Almost 200GB per 24h)

The problem with the increasing file size also remains so the transfer rates are actually even slower since I calculated the transfer rate with the file size given by Airbyte and not the actual size of the CSV initially on S3.

It would be great if someone could comment on this because I think the bottleneck here is the Airbyte software design. Are there currently any efforts to make this more efficient?