s-Server and the Linear Road Benchmark

The Linear Road Benchmark is a well-known measure for assigning an L-rating to stream processing platforms. An L-rating is a number of highways for which traffic analytics can be performed on a given hardware system. It is described here:

http://www.cs.brandeis.edu/~linearroad/

The Linear Road Benchmark assumes a fictional metropolis that measures 100 x 100 miles with the following characteristics:

  • The city has 10 Expressways every 10 miles.
  • At every mile mark, each each expressway has an exit ramp and an on ramp.
  • Each expressway has 4 lanes in each direction: 3 travel lanes and one lane for entrance and exit.

Traffic Data:

On the Linear road network of the Linear city as described above

  • Every vehicle emits a position report every 30 seconds.
  • One accident occurs randomly on each expressway every 20 minutes, taking 10 to 20 minutes to clear.

The test measures four types of events:

  • Type 0 (Report): Accident notification, assuming that an accident occurs when two vehicles are stopped. In SQLstream’s test, 99% of events are real-time position reports.
  • Type 2 (Request): Requests for account balances for each vehicle. It returns a total account amount for any driver that requests it.
  • Type 3 (Request): Daily tolls expenditure for a specific day in the past 10 weeks.
  • Type 4(Request): Travel time predictions for a journey on a given expressway, day of the week and time of day.

For all request types above, the requirement was to respond with up to a maximum of specified latency.

SQLstream results:

Technology Hardware Nodes 80 Core L-Rating
Guavus SQLstream MacBook Pro 2013 VM: 4 Cores, 8 GB RAM 1 1200 (60 on 4 cores with linear scale out)
IBM Streams Four Azure A11 nodes, each node has: 16 cores, 112 GB RAM, 382 GB Disk, 10 Gbit/s networking, or CPU model: 45, Intel®, Xeon®, CPU E5-2670 0 @ 2.60 GHz 6 (4 for stream processing and 2 for ingest) 200
Apache Apex Four Azure A11 nodes, each node has: 16 cores, 112 GB RAM, 382 GB Disk, 10 Gbit/s networking CPU model: 45, Intel®, Xeon®, CPU E5-2670 0 @ 2.60 GHz 6 (4 for stream processing and 2 for ingest) 102
Apache Storm Four Azure D14, each node has: 16 cores, 112 GB RAM, 800 GB Disk (SSD), 1 Gbit/s) CPU model: 45, Intel®, Xeon®, CPU E5-2670 0 @ 2.20 GHz 6 (4 for stream processing and 2 for ingest) 10

Please contact SQLstream at sales@sqlstream.com to conduct the linear road benchmark.