On disk index scan has been implemented:

  • It is now possible not to load index files of the last fusion stage in RAM.

    This enhancement allows to reduce memory usage several times with almost no query performance degradation.

Full scan queries became twice as fast due to block level parallelization enhancement.

Key parameter queries performance enhanced for up to 20% due to decompression optimization.

Predicates priority has been reviewed to ensure optimal query execution planning.

Memory leak related to JMX interface has been fixed.

Parallel search issue in the sharded mode has been fixed.

The query syntax has been redesigned to make it stricter and more similar to other NoSQL storages.

  • The following query predicates are now supported:

    • $eq - equal
    • $gt, $gte - greater / greater or equal
    • $lt, $lte - less / less or equal
    • $in - equal to any value in the array
    • $regex - matches PCRE regular expression
  • Custom wildcard implementation has been replaced with common regex syntax.

Second level sorting has been implemented:

  • It is now possible to additionally sort by any column inside second level shards / sst blocks. This enhancement has improved compression for up to 30% without query performance penalty.

Convinient administration utility has been introduced, allowing to:

  • access different information: table schemas, running queries, storage statistics, version and license
  • add new table
  • manually run and stop queries
  • munually launch loading, fusion and cleaning
  • schedule old data removal
  • update license

JMX statistics interface has been implemented.

Directory storing loaded raw files can now be cleaned by timeout.

Memory leaks during initial loading and merging file processes have been fixed.

Minor compression bug fixes.

Memory leak related to inaccurate index memory storage has been fixed.

The initial loading process became more efficient and up to 45% faster.

Memory usage during the fusion process has been reduced for up to 60%.

Full scan batch queries performance has been increased for queries with a big number (several thousands or more) of non-key parameters.

Http interface for old data removal has been improved.

The new version supports JMX technology, which allows easy Equator Server monitoring.

Logging information became more structured and informative.

Http interface for old data removal has been introduced.

Internal parallelism has been improved:

  • Full scan queries can now be parallelized on block level for each sharded/sstable file.

Loading from compressed files has been implemented:

  • Data can now be loaded from gzip/snappy compressed files. The algorithm is automatically detected based on file extension.

Http interface for old data removal has been introduced.

Virtual fields feature has been introduced:

  • This feature allows to specify a virtual field in table schema as a operation (e.g. sum), applied to regular table fields. The virtual field can be then used in "collectMinMax" schema section and in the query conditions.

List queries feature has been introduced:

  • For any one field in the search conditions a list of values can be specified instead of a single value.

Equator Server is the new generation CopperEye product:

  • Designed to satisfy the need of telecommunication providers for big data retention, Equator makes it possible to query petabytes in seconds.

Two storage engines are introduced - sstable and sharded:

  • Those engines utilize modern approaches to data warehousing without using indexes and make it possible to find an optimal balance between query performance and storage density. Such technologies as partitioning, binary data representation,intelligent compression and bloom filters are also used.

Faster search and loading, denser storage:

  • Equator queries are up to 100 times faster than those in Greenwich, data is loaded up to 10 times faster and utilizes up to 6 times less disk space.

Convenient interfaces:

  • HTTP JSON and JDBC APIs are implemented for easy integration with third party applications.

Data Acquisition Bus:

  • A modern replacement to CopperEye CSync is introduced, designed to validate, transform and deliver data to Equator cluster.