Quick Requirements Run-through

Production Environments

  • 32 GB minimum RAM recommended
  • Minimum free disk space depends on project requirements, should be above dev environment requirement for disk space

Development and Testing Environments

  • 8 GB minimum RAM recommended
  • 10 GB minimum free disk space recommended


Hardware

Basic Aspire has a fairly small footprint. Dev/Test installs can be done on any modern (<5 years of age) mid-high spec laptop. For most production implementations, we recommend:

  • A server with >=4 physical core processor running a 64-bit Linux or Windows instance with Java 11 SE 64-bit properly configured.
  • JVM Heap Size set to a minimum of 8 GB RAM (32 GB recommended for production).
    • Heap Size allocation can be modified at startup, please change corresponding heap settings in <aspire install path>/bin/aspire.bat or <aspire install path>/bin/aspire.sh.

    • For Aspire instanced installed as a service, please check the guides for Run as a Windows Service or Run as a Linux Service.

    • 8 GB is the minimum recommended heap size for lightweight Aspire instances. This may change depending of the content sources and the data to process.
  • Minimal disk space: The Aspire binaries and dependencies take up less than 1GB; however we recommend a minimum of 10 GB free disk space for development and testing environments
    • We don't recommend the minimal disk space for production environments - we've observed production environments that easily occupy scores of GB of disk space

    • Logs and data generated from crawls and other component activity will occupy additional disk space aside from what's mentioned above. Please plan your storage needs accordingly.

  • Network connection speeds of 100 Mbps minimum, preferably 1 Gbps or 10Gbps. If using an online instance of Aspire, a broadband/fiber connection with low latency is preferred for quicker Aspire component updates.

Factors Affecting Recommendations

What you need to run Aspire optimally in a complex, enterprise production environment depends on multiple factors, including:

  • The total number of data sources to be crawled, and for each data source:
    • What type of repository holds the data (i.e. SharePoint, Documentum, RDBMS, etc.)
    • Total number of documents
    • Total number of bytes
    • How long an initial indexing crawl should take
    • How long a periodic incremental indexing crawl should take
    • Are real-time updates required?
    • Network speed between repository and Aspire server
  • Data metrics for query client connections:
    • Number of client connections expected
    • Number of queries per second expected
    • Query latency times required / expected
    • Is authentication, authorization, and group expansion required for client search requests?
  • Will distributed processing be required?

Please feel free to contact your Accenture Search and Content Analytics team representative for further assistance in setting up your Enterprise Aspire environment. We're more than happy to be of service to you.


Network

Aspire makes substantial use of Maven, in both the build and execution phase. During the build process, Maven will download bundles (jar files) and plugins from several Maven repositories located on the internet (including, but not limited to https://repository.searchtechnologies.com/artifactory/public and http://repo1.maven.org/maven2). During execution, Aspire will download applications and components (OSGI bundles or jar files) from https://repository.searchtechnologies.com/artifactory/public.

If allowing Aspire access to the internet is not in line with your network policy, see Setting Aspire with No Internet Access (Offline) or Using your own Maven repository with Aspire.

Another network consideration is the speed of the network between the content repository and the Aspire server. When an initial crawl or incremental crawl is performed it is typically fetching the entire document and associated metadata. Although the text of each document may be rather small, the full binary can be quite large and if the actual network speed, particularly for WANs, is slow, the overall performance of the system will be slow. There are solution architectures that can be employed to address this situation. Please feel free to contact your Accenture Search and Content Analytics team representative for further assistance and we would be happy to discuss these issues and potential solutions.

  • No labels