I'm shopping for a server that will allow my coworkers to use QGIS through a terminal server. We have 2 Tb of orthophotos in tif, twf, ecw and shapefiles format. They tell me that actually when they work with orthophotos it can be sluggish. The way they describe it, the loading time of the orthophotos seems to be the culprit. They work with the orthophotos stored on an external HDD through an USB 2.0 connection. I consider buying extra big SSD drives to store the orthophotos on them to assure the best performance, however an IT consultant tells me he thinks it would be relevant and the performance gain compared to using HDDs for those file's storage would not be significant (we will however put the OS and the main files on SSD drives). Has anyone experienced working with orthoimages on SSD drives, and is the performance gain significant compared to working with orthoimages on a HDD ?

  • related: gis.stackexchange.com/questions/22623/… – Mapperz Dec 7 '15 at 18:37
  • 2
    Your USB 2.0 connection is the bottleneck here. Max throughput of ~35MB/s. If SSD is too pricey, look into a good enterprise internal HDD over SAS or SATA 3. – Paul Dec 7 '15 at 18:46
  • 1
    Difference between rotating disks and SSD has meaning only after everything else is done right. With ECW there is not much to adjust but for tiff files it is essential to have them with internal tiling and with overviews. The GeoTIFF part of slideshare.net/geosolutions/geoserver-on-steroids-foss4g-2015 is good basis for all server marks, not only for GeoServer. QGIS is desktop but same recipes suit for it too. – user30184 Dec 7 '15 at 18:53

Read/write speeds should be much better if the orthophotos were stored on an internal HDD rather than external. This would also be considerably more cost-effective than buying a large volume SSD.



I'd say the main problem here may be that you have multiple users accessing photos on a single USB hard drive. The USB connection limits the potential transfer speed as Paul points out, but if your users are using the drive simultaneously, the transfer speed of the drive becomes less important and its response time becomes more important. Of course, an SSD beats an HDD handily on both transfer speed and response time, response times on HDDs are measured in milliseconds, on an SSD response times are < 1ms. In my experience, everything is faster in GIS when your data is stored on an SSD. The larger the dataset, the greated the improvement in speed over an HDD (even a RAID array of HDDs on a server).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.