This lists some possible improvements to Vespa which have been considered or requested, can be developed relatively independently of other work, and are not yet under development. For more information on the code structure in Vespa, see Code-map.md.
Effort: Low
Difficulty: Low
Skills: Java, C++, multithreading
Currently, trace information can be requested for a given query by adding travelevel=N to the query. This is useful for debugging as well as understanding performance bottlenecks. However, the trace information only includes execution in the container, not in the content nodes. This is to implement similar tracing capabilities in the search core and integrating trace information from each content node into the container level trace. This would make it easier to understand the execution and performance consequences of various query expressions.
Code pointers:
Effort: Low
Difficulty: Low
Skills: Java
Query profiles make it simple to support multiple buckets, behavior profiles for different use cases etc by providing bundles of parameters accessible to Searchers processing queries. Writes go through a similar chain of processors - Document Processors, but have no equivalent support for parametrization. This is to allow configuration of document processor profiles by reusing the query profile support also for document processors.
Code pointers:
Effort: Medium
Difficulty: Low
Skills: Java
Some times there is a need to reindex existing data to refresh the set of tokens produced from the raw text: Some search definition changes impacts the tokens produced, and changing versions of linguistics libraries also cause token changes. As content clusters store the raw data of documents it should be possible to reindex locally inside clusters in the background. However, today this is not supported and content need to be rewritten from the outside to refresh tokens, which is inconvenient and suboptimal. This is to support (scheduled or triggered) background reindexing from local data. This can be achieved by configuring a message bus route which feeds content from a cluster back to itself through the indexing container cluster and triggering a visiting job using this route.
Code pointers:
- Document API which can be used to receive a dumpt of documents: DocumentAccess
Effort: Medium
Difficulty: Low
Skills: Java, C++, networking
Currently, search requests happens over a very old custom protocol called "fnet". While this is efficient, it is hard to extend. We want to replace it by RPC calls. An RPC alternative is already implemented for summary fetch requests, but not for search requests. The largest part of this work is to encode the Query object as a Slime structure in Java and decode that structure in C++.
Code pointers:
- FS4 protocol search invokers (to be replaced by RPC): FS4SearchInvoker, FS4FillInvoker
- Current Query encoding (to be replaced by Slime): QueryPacket
- Slime: Java, C++
- C++ query (to be constructed from Slime)
Effort: Medium
Difficulty: Low
Skills: Java
There is currently support for creating Application instances programmatically in Java to unit test application package functionality (see com.yahoo.application.Application). However, only Java component functionality can be tested in this way as the content layer is not available, being implemented in C++. A Java implementation, of some or all of the functionality would enable developers to do more testing locally within their IDE. This is medium effort because performance is not a concern and some components, such as ranking expressions and features are already available as libraries (see the searchlib module).
Code pointers:
- Content cluster mock in Java (currently empy): ContentCluster
- The model of a search definition this must consume config from: Search
Effort: Medium
Difficulty: Medium
Skills: Java, C++, distributed systems
Support "update where" operations which changes/removes all documents matching some document selection expression. This entails adding a new document API operation and probably supporting continuations similar to visiting.
Effort: Medium
Difficulty: Medium
Skills: C++, multithreading, performance, indexing, data structures
Vespa supports maps and and making them searchable in memory by declaring as an attribute. However, maps cannot be indexed as text-search disk indexes.
Code pointers:
Effort: High
Difficulty: High
Skills: C++, Java, distributed systems, performance, multithreading, network, distributed consistency
Vespa instances distribute data automatically within clusters, but these clusters are meant to consist of co-located machines - the distribution algorithm is not suitable for global distribution across datacenters because it cannot seamlessly tolerate datacenter-wide outages and does not attempt to minimize bandwith usage between datacenters. Application usually achieve global precense instead by setting up multiple independent instances in different datacenters and write to all in parallel. This is robust and works well on average, but puts additional burden on applications to achieve cross-datacenter data consistency on datacenter failures, and does not enable automatic data recovery across datacenters, such that data redundancy is effectively required within each datacenter. This is fine in most cases, but not in the case where storage space drives cost and intermittent loss of data coverage (completeness as seen from queries) is tolerable.
A solution should sustain current write rates (tens of thousands of writes per ndoe per second), sustain write and read rates on loss of connectivity to one (any) data center, re-establish global data consistency when a lost datacenter is recovered and support some degree of tradeoff between consistency and operation latency (although the exact modes to be supported is part of the design and analysis needed).
Code pointers:
Effort: High
Difficulty: High
Skills: Java, C++, distributed systems, performance, networking, distributed consistency
Tensors in ranking models may either be passed with the query, be part of the document or be configured as part of the application package (global tensors). This is fine for many kinds of models but does not support the case of really large tensors (which barely fit in memory) and/or dynamically changing tensors (online learning of global models). These use cases require support for global tensors (tensors available locally on all content nodes during execution but not sent with the query or residing in documents) which are not configured as part of the application package but which are written independently and dynamically updateable at a high write rate. To support this at large scale, with a high write rate, we need a small cluster of nodes storing the source of truth of the global tensor and which have perfect consistency. This in turn must push updates to all content nodes in a best effort fashion given a fixed bandwith budget, such that query execution and document write traffic is prioritized over ensuring perfect consistency of global model updates.
Code pointers: