Java API services very slow against IDR

[dominikl]

See Omero Java API with public data hosted on the IDR server .

I can confirm, using the BrowseFacility of the gateway it takes nearly 30sec. Same using the ContainerService of the Java API, which is expected as that's what the BrowseFacility is using under the hood. In contrast to < 100ms using the QueryService!

It's not really an issue of the Java Gateway but rather of the Java API ContainerService, might have to move the issue to a different github repository.

Example code

try (Gateway g = new Gateway(new SimpleLogger())) {
            long datasetId = 7808;  // idr0066 experimentD: https://idr.openmicroscopy.org/webclient/?show=project-854 
            ExperimenterData exp = g.connect(loginCredentials);
            SecurityContext ctx = new SecurityContext(exp.getGroupId());

            ParametersI param;
            DatasetI tmp;
            DatasetData dataset;
            long start;

            System.out.println("BrowseFacility");
            start = System.currentTimeMillis();
            Collection<ImageData> imgs = g.getFacility(BrowseFacility.class).getImagesForDatasets(ctx, Collections.singletonList(datasetId));
            for (ImageData img : imgs)
                System.out.println(img.getName());
            System.out.println(System.currentTimeMillis() - start);

            System.out.println("ContainerService");
            IContainerPrx cs = g.getPojosService(ctx);
            start = System.currentTimeMillis();
            param = new ParametersI();
            param.leaves();
            tmp = (DatasetI) cs.loadContainerHierarchy("Dataset", Collections.singletonList(datasetId), param).iterator().next();
            dataset = new DatasetData(tmp);
            for (Object img : dataset.getImages())
                System.out.println(((ImageData)img).getName());
            System.out.println(System.currentTimeMillis() - start);

            System.out.println("QueryService");
            IQueryPrx qs = g.getQueryService(ctx);
            param = new ParametersI();
            param.addId(datasetId);
            String query = "select d from Dataset as d " +
                    "join fetch d.imageLinks as l " +
                    "join fetch l.child as i " +
                    "where d.id = :id";
            start = System.currentTimeMillis();
            tmp = (DatasetI) qs.findByQuery(query, param);
            dataset = new DatasetData(tmp);
            for (Object img : dataset.getImages())
                System.out.println(((ImageData)img).getName());
            System.out.println(System.currentTimeMillis() - start);
        }

Output

BrowseFacility
ExpD_chicken_embryo_MIP.ome.tif
ExpD_chicken_embryo_stitched.ome.tif
29038
ContainerService
ExpD_chicken_embryo_MIP.ome.tif
ExpD_chicken_embryo_stitched.ome.tif
28236
QueryService
ExpD_chicken_embryo_stitched.ome.tif
ExpD_chicken_embryo_MIP.ome.tif
66

[imagesc-bot]

This issue has been mentioned on Image.sc Forum. There might be relevant details there:

https://forum.image.sc/t/omero-java-api-with-public-data-hosted-on-the-idr-server/73946/2

[jburel]

This is not surprising since the amount of data that is loaded is different
See https://github.com/ome/omero-server/blob/master/src/main/java/ome/services/query/PojosLoadHierarchyQueryDefinition.java

One option will be to offer "light" queries from the gateway itself

[dominikl]

One option will be to offer "light" queries from the gateway itself

Good idea.

[dominikl]

That's actually covered by above mentioned PR.