4

I have installed PostgreSQL 9.3.5 and PostGIS 2.1.2 on a 16Gb Ubuntu 14.04 system.

  1. If I load all the states' TIGER data, using the scripts provided in Postgis, how big will the database be for the entire US?

    I have currently installed just the TIGER data for California (CA).

  2. It takes about a minute to geocode an address. I have read the performance hints suggested at How to improve poor performance times for PostGIS 2.1 Tiger Geocoder?, and have tried setting the shared memory buffer to a much larger value. I haven't been able to improve the query performance time.

Do you have suggestions?

Vince
  • 20,017
  • 15
  • 45
  • 64
Dan Connors
  • 41
  • 2
  • Dan, did you ever get a clear answer to your question? I just hit 100GB loading the states alphabetically when I got to "wy" so it's probably around 100GB. I'm moving data around and rebuilding indexes now but can answer at least that part of the question in a little bit. My guess is that it's going to be 100-120GB. – aaryno Jun 11 '15 at 15:22
  • Can you please share your server details like RAM size and other configurations? – Sonil Shrivastava Jun 28 '17 at 13:59

2 Answers2

2

  1. Total size for the data will be 105G. However you need some temp space for downloading and provisioning the database. I used to use a 120G SSD and it was filled up before finish. I assigned 180G to it in second time without problem. You probably need at least 105G + 30G, since my download folder after the job finish still have 30G files in it.

  2. 1 min is not right and usually your problem is not in database config, since the default config should at least give you 300 ms. I'd suggest these directions:

    • Install the database in SSD. It make huge difference.
    • Format the address correctly, just like the pprint function output.

I cannot give more suggestions without more information of your setup and queries. You can check my blog post about my setup and my script.

dracodoc
  • 370
  • 4
  • 14
0
  1. dracodoc has a good answer for you for this aspect of your question
  2. 1 minute for a geocode query indicates that you haven't done step 15 of section 2.4.1 here
Nate Symer
  • 101
  • 1