10/22/2018: New Features

Support of Gurobi Optimizer 8.1

The version 8.1 is now supported in addition to 8.0.1. Clients can start upgrading to the latest version.

Support of Microsoft Azure (v8.x engine only)

Gurobi Instant Cloud now supports machines in three Microsoft Azure regions: US East, US West 2 and EU West. You can choose between the following machine types:

  • Compute Optimized: F16v2, F32v2, F72v2 machines with 8, 16 and 36 physical cores respectively and up to 144GB of memory. The F-series is hyper-threaded and based on the 2.7 GHz Intel Xeon® Platinum 8168 (SkyLake) processor, which can achieve clock speeds as high as 3.7 GHz with the Intel Turbo Boost Technology 2.0.
  • High Performance Compute: H16m machine with a configuration of 16 physical cores and 224GB of memory. This machine is based on the Intel Xeon E5-2667 v3 Haswell 3.6 GHz with turbo always on. On this machine, hyper-threading is not enabled so that cores can better support high performance computing workloads. Note that this machine may not be available in all regions.
  • Memory Optimized: E64v3 machine with a configuration of 32 physical cores and 432GB of memory. This machine is based on the 2.3 GHz Intel XEON® E5-2673 v4 (Broadwell) processor and can achieve 3.5 GHz with Intel Turbo Boost Technology 2.0.

Support of new AWS machine types c5, r5 and z1

You can now choose between the following machine types:

  • Compute Optimized: c5.4xlarge, c5.9xlarge, c5.18xlarge machines with 8, 18, 36 physical cores respectively and up to 144GB of memory. These machines are based on the Intel® Xeon® Platinum 8124M CPU @ 3.00GHz, and allow a single core to run up to 3.5 GHz using Intel Turbo Boost Technology.
  • High Performance Compute: z1d.12xlarge machine with 24 physical cores and 384GB of memory. This machine is based on a custom Intel® Xeon® Platinum 8151 CPU delivering a sustained all core frequency of up to 4.0 GHz. Note that this machine may not be available in all regions.
  • Memory Optimized:
    • r5.24xlarge with 48 cores and 768GB of memory using the Intel Xeon Platinum 8000 series processors with a sustained all core frequency of up to 3.1 GHz. Note that this machine may not be available in all regions.
    • r4.8xlarge and r4.16xlarge machines with 16 or 32 physical cores and up to 488GB of memory. These machines are based on the Intel® Xeon® CPU E5-2686 v4 @ 2.30GHz.

The previous generation of c4 and r3 machines have been upgraded to c5 and r4 machines. Instant Cloud pools and preferences have already been changed so that you will benefit from the new machines immediately.

4/24/2018: New Features for the v8 Engine Only

Better Security and Easier Integration with HTTPS

With the v8 engine, all the communications between the clients and the Gurobi Instant Cloud are using the HTTPS protocol. This means that your communications are secured and encrypted using standard internet protocols. In addition, Gurobi servers enforce the latest encryption policies (TLS v1.2 and above only). For better security, the dedicated machines started by Instant Cloud on your behalf cannot be accessed directly, and all communications must transit through a secured and highly-available region router acting as a reverse proxy. This also facilitates the integration with clients as only the standard HTTPS protocol and standard port 443 need to be open if a firewall is in place.

For customers who have used versions v6.x and v7.x, the new v8 engine will simplify their network integration, as there is no need to open ports 61000 to 65000 anymore. Also, please note that the pool fixed IP feature of v6.x and v7.x engines is not available for v8 as we have replaced the protocols by internet standards.

Active Job List

The v8 engine now provides visibility into the jobs running on the machines. The Instant Cloud Manager has a new menu option to list the jobs currently running or queued on all active machines and pools. The list also contains recently ended jobs with a status:

  • completed: the job ended successfully
  • aborted: the job was aborted
  • disconnected: the connection with the client was lost because the client terminated or experienced a network issue
  • idle timeout: the client was still connected but was idle for too long (see IdleJobTimeout)
  • failed: the job running on the server experienced an unexpected problem and failed, for example the job went out of memory.

Note that when a machine is terminated, the job executed on this machine will disappear from this list. If you need to keep visibility on all executed jobs, see the Job History feature below.

It is also possible to select jobs and abort them.

Active Job Dashboard

From the active job list, the detailed job dashboard can de displayed. The job dashboard provides different views about a given job in order to let an optimization expert monitor the job execution:

  • general job information and status
  • current KPIs of the model and the algorithm
  • changed parameters passed to the job
  • charts showing the progress of the optimization
  • live engine log

Job History

While the active job list provides a view of jobs currently queued, running or recently ended, the job history provides a persistent list of ended jobs. The new job history menu option lets you browse and filter these jobs. The engine log can also be downloaded. When using this feature, the job metadata and the engine logs are stored on the Instant Cloud secured servers for a maximum of 90 days. This feature can be disabled on selected pools or machines in the options panel.

Pool Scaling

It is now possible to scale up or down your pools using the Instant Cloud Manager or the REST API. A pool defines a number of compute servers that is actually the minimum number of servers. So when a client starts a pool, the pool will be ready when this number of servers is ready. Then, you can scale up by requesting additional servers to be added to the pool. When servers are added, they automatically join the cluster of compute servers of the pool and start processing new jobs or jobs already queued in existing servers.

The pool will scale down automatically by using the existing idle shutdown parameter. This means that any server being idle for this time limit will be terminated, up to completely terminate the pool. The pool can also be explicitly scaled down by reducing the number of servers. In this case, if a machine must be stopped but it is already running a job, it will be moved into a draining mode where new jobs will not be processed while running ones will be processed normally. When these jobs have completed, the machine will be automatically terminated.

For better monitoring, the pool view displays the number of machines that are starting, stopping and the ones that are ready.

Machine Metrics

The machines are now reporting the current CPU and memory usage. These metrics are displayed in the machine view.

Idle Job Timeout

A parameter has been added to specify the duration limit after which the job will auto-terminate when there is no command sent to the server. This is useful to avoid using resources when some clients may leave their connections open (for example in an interactive python shell) while not being active. Clients can set this timeout using the CSIDLETIMEOUT parameter or using the IDLETIMEOUT property in the license file. The default value is 30 minutes. The pool or machine option can be used to set a larger timeout globally that will override the client default.

Extended REST API

The REST API has several new endpoints to support the monitoring of active jobs, the access to job history and the management of pool scaling. See the REST API reference documentation for more details.


4/24/2018: Global Updates

Supported Regions

In order to provide better support on heavily used regions, we have decided to focus our resources on 5 regions:

  • us-east-1 (N. Virginia)
  • us-west-1 (N. California)
  • eu-central-1 (Frankfurt)
  • ap-northeast-1 (Tokyo)
  • ap-southeast-2 (Sydney)

As a consequence, we are discontinuing the support of 3 regions and if you had pools using these regions, they will be automatically moved to the nearest one as follows:

  • eu-west-1 (Ireland) is replaced by eu-central-1 (Frankfurt)
  • us-west-2 (Oregon) is replaced by us-west-1 (N. California)
  • ap-southeast-1 (Singapore) is replaced by ap-southeast-2 (Sydney)

Supported Engine Versions

As we are now supporting the new v8 engine, we have removed the support of v6.0 and v6.5 in Instant Cloud. Our policy is to support the last two major versions, namely v8 and v7.x. If you had pools and preferences using these versions, they will be automatically upgraded to v7.0.


7/12/2017: Added Support for the v7.5 Engine

The new v7.5 engine can now be selected in the dropdown menu of the engine versions of your pools and machines. Note that the version of the client and the machines or pools must match. If you had existing pools, they will still use the current version. When you are ready to use the v7.5, you will need to manually update the version of your pools to match the client so that you can upgrade at your own pace.