New in InterSystems IRIS for Health 2024.3

A set of enhancements to the FHIR Server Search capabilities have been introduced, including:

• _maxresult is now supported for FHIR Search

• :missing modifier is now supported for FHIR Search

• :not modifier is now supported for FHIR Search

• Value-quantity search performance has been enhanced

JSON Advanced SQL (JSONAdvSQL) strategy search performance enhancement of the following:

• Search for patient by date of birth

• Search for observations by quantity

• Search for patient by patient ID

This features introduces tracking for various interoperability metrics including:

• Message Volume: Tracks the volume of messages moving through the system to provide operational insights.

• Database Impact of Messages: Measures how message traffic affects database performance.

• System Usage Tracking: Monitors real-world usage of interoperability features to support operational decision-making.

A set of enhancements have been introduced to streamline the usability of the HL7 Production Validator:

• Simpler Production Comparison Workflow: Simplifies and reduces the steps requires to compare the legacy and upgraded productions.

• HL7 Segment Skipping: Option for automatic skip of empty HL7 segments to comparison view to focus on real differences in the message structure.

• Service/Operation Skipping: Option for automatic skip of a specified Service/Operation pairing.

• Backward Compatibility: There is now support for backward compatibility with Caché 2016.x versions as a legacy system.

A new system database IRISMETRICS has been introduced for storing interoperability usage metrics. No customer application should directly interact with the IRISMETRICS database, which exists purely for internal use by InterSystems IRIS for Health.

This feature introduces a simple “Resend” button in Visual Trace, making it easier for users to resend messages directly from the trace without performing complex message searches. This significantly reduces the time required to correct message flows in integration scenarios.

Several enhancements have been introduced to the rule editor, improving its usability and scalability:

• Add Filter by Rules: Adds a toolbar filter to simplify rule viewing and editing.

• Auto Scroll to Specific Rule: Automatically scrolls to a specific rule within the editor.

• Support for Large Rule Sets: Enhances the ability to handle and manage large rule sets.

This release significantly extends the Vector Search capability introduced with InterSystems IRIS 2024.1.0. Optimization of the low-level vector similarity operations and a more tailored storage model for vector data has yielded a 3-4x speedup of raw vector search speed.

This release also introduces a new EMBEDDING datatype that automates the creation of vector embeddings for text or other content captured in another column. This removes the need for user code to invoke an embeddings model and pass those vector embeddings in manually when ingesting new rows. An EMBEDDING column simply refers to a source column in the same table, and a named configuration that defines the specific embeddings model or service to use.

For more about these new capabilities, see Using Vector Search.

InterSystems IRIS Adaptive Analytics version 2024.1.3 includes AtScale version 2024.1.3 and an updated User Defined Aggregate Function (UDAF) file. This release includes the following new modeling and BI capabilities:

• Formatting of calculation output based on the field input rather than just the format defined on the calculation. This is particularly useful for calculations that cannot have a standard output format, such as year-over-year growth in dollars, euros, or time.

• Support for the following MDX functions: DatesPeriodsToDate, DatesMTD, DatesQTD, DatesWTD, DatesYTD.

• Support for Microsoft Excel’s custom pivot table dimension grouping functionality. This enables you to create custom groups within Excel, without having to make changes to your models. This change also includes support for session sub-cubes, naming, totaling, filtering, and ungrouping.

For more detail on AtScale 2024.1.3, see the AtScale release notesOpens in a new tab.

For more information about Adaptive Analytics, refer to the InterSystems documentation and online learningOpens in a new tab content.

SQL query plans now may include informational and warning messages to inform users about possible opportunities to improve the query structure, schema, or system settings. Messages may include warnings about tables lacking statistics, indices that are marked as non-selectable, collation mismatches, and more. Future releases will expand the types of messages users may see, and include mechanisms to easily aggregate and consume this information for at the namespace level.

The WITH clause, used to define Common Table Expressions (CTEs) at the beginning of a statement, can now also be used in DML statements. This means complex INSERT ... SELECT  statements, or UPDATE and DELETE statements with a complex FROM clause, can now be simplified with upfront CTE definitions, increasing the overall readability of the statement.

Administrators can now define more fine-grained auditing policies for SQL statements, refining them based on statement type in addition to the access mechanism (database driver, Dynamic SQL, or Embedded SQL) as was available before. For example, a policy may audit all statements that modify data (DML statements), but ignore queries that only read data.

On Linux and Unix platforms, InterSystems IRIS will now use fast storage allocation functionality when available in the underlying filesystem to expand databases and WIJ files. Previously, database expansion happened through writing empty blocks, whereas the new approach involves a metadata operation only, improving the speed of database and WIJ file expansion as much as 50 to 100-fold. The filesystems for which this new approach is supported are XFS, ext4, Btrfs on Linux, and JSF2 on AIX. Other platforms and filesystems may see limited improvements thanks to other optimizations to the expansion code introduced in this release.

Starting with this release, users can convert a non-sharded table to a sharded table and vice versa, even when data is present in the table. Using a new ALTER TABLE ... CONVERT command, you can change not only the table’s storage definition, but also distribute data from the shard master database across shards (or vice versa). This is an online operations, and applications and queries can continue to read from and write to the table while conversion is running. API methods are available to check progress for this potentially long-running operation.

When using Columnar Storage to achieve top analytical query response times, data is organized in $vector chunks holding up to 64000 values at a time. For string fields with high cardinality (many distinct values), these data structures risked hitting lower level limitations on what can be stored in a single unit, and we automatically reverted the storage mode of such columns to row layout based on the column's defined string length. With this release, we're introducing a mechanism to automatically adjust the chunk size based on the column's defined length, such that longer string sizes can still be treated using columnar storage and benefit from fast response times when used in analytical queries.

InterSystems has provided two flavors of kits for AIX users this year: one that supports OpenSSL 1.x and one that supports OpenSSL 3.x. Starting with the current 2024.3 release, InterSystems will only provide kits that support OpenSSL 3.x.