Cassandra cql: how to select the LAST n rows from a table

Question

I want to verify that rows are getting added to the table. What cql statement would show the last n rows from the table below?

Table description below:

cqlsh:timeseries> describe table option_data;

CREATE TABLE option_data (
  ts bigint,
  id text,
  strike decimal,
  callask decimal,
  callbid decimal,
  maturity timestamp,
  putask decimal,
  putbid decimal,
  PRIMARY KEY ((ts), id, strike)
) WITH
  bloom_filter_fp_chance=0.010000 AND
  caching='KEYS_ONLY' AND
  comment='' AND
  dclocal_read_repair_chance=0.100000 AND
  gc_grace_seconds=864000 AND
  index_interval=128 AND
  read_repair_chance=0.000000 AND
  replicate_on_write='true' AND
  populate_io_cache_on_flush='false' AND
  default_time_to_live=0 AND
  speculative_retry='99.0PERCENTILE' AND
  memtable_flush_period_in_ms=0 AND
  compaction={'class': 'SizeTieredCompactionStrategy'} AND
  compression={'sstable_compression': 'LZ4Compressor'};

cqlsh:timeseries>

Spondylitis

Answer 1

You didn't specify last n "by what".

To get the last N per id:

SELECT * FROM option_data WHERE ts=1 ORDER BY id DESC LIMIT N;

ORDER BY clause can only be applied to the second column in a compound primary key. If you need to query by time you will need to think about your data model a little more.

If your queries are most often "last N", you might consider writing something like this:

CREATE TABLE time_series (
    id text,
    t timeuuid,
    data text,
    PRIMARY KEY (id, t)
) WITH CLUSTERING ORDER BY (t DESC)

... where 'id' is your time series id. The CLUSTERING ORDER reverses the order of timeuuid 't', causing the cells to be stored in a natural order for your query.

With this, you would get the last five events as follows:

SELECT * FROM time_series WHERE id='stream id' LIMIT 5;

There is a lot of information out there for time series in Cassandra. I suggest reading some of the more recent articles on the matter.

Lobelia

the link is dead already, I found this helpful – Vesiculate