Univa solution simplifies hybrid and multi-cloud Slurm deployments
HPC cloud is on the rise
While enterprise HPC users rely on on-premise infrastructure for most workloads, cloud spending is on the rise. According to Hyperion, 74% of HPC users run at least some workloads in the cloud, and HPC cloud usage is expected to expand from just 10% of workloads today to 15% by 2021, outpacing the growth of the HPC market as a whole(1). Cloud offerings are increasingly capable, and while they can be more costly than on-premise alternatives, cloud users avoid the hassle of installing and managing local infrastructure and benefit from fast access to specialized resources and flexible pay-per-use consumption models.
Slurm is a popular choice
It will be no surprise to HPC users that open-source Slurm (the Simple Linux Utility for Resource Management) is a popular HPC workload manager. Slurm is well known and is used by approximately 60% of the world’s Top 500 supercomputers(2).
What may come as a surprise, however, is the use of Slurm in the cloud. In a recent survey of HPC users sponsored by Univa, 45% of HPC users reported using Slurm, and all of these respondents ran at least some workloads in the cloud. Some Slurm users reported spending over USD 250K per month on cloud services(3). In the same survey, 70% of Slurm users reported that HPC cloud is “definitely a winner” or that they see “real potential.”
Cost management is a major challenge
Gartner estimates that 80% of cloud IaaS users will overshoot their budgets through 2020 because they lack necessary process controls to deal with costs in the cloud(4). Also, Flexera’s RightScale 2019 State of the Cloud Report finds that as much as 35% of cloud spending is wasted(5). The ease with which cloud resources can be provisioned is a big part of the problem. Users often overshoot cloud budgets because they fail to shut down idle or orphaned instances, deploy more resources than necessary, or fail to manage compute and storage resources optimally.
While budgeting for cloud expenses seems straightforward, it is surprisingly complicated. Most cloud providers offer multiple rate-structures – as examples, reserved instances, on-demand instances, spot instances, spot fleets, and costs can vary by region. Also, while machine instances account for the majority cloud spending, users consume object and block storage, file systems, and network-related services such as VPCs, gateways, and VPNs. Each service has its own pricing model, often with multiple chargeable components.
Multi-cloud adds to the complexity
The rise of multi-cloud deployments is adding to the complexity. Despite the best efforts of HPC managers to standardize on a single cloud provider, Gartner indicates that 80% of cloud users already deal with more than one provider(6). Multi-cloud environments inevitably occur because of third-party collaborations, M&A activities, LOBs making independent purchasing decisions, managed service providers, and SaaS and PaaS solutions with affinities to particular clouds.
Cloud-spend association is key
A key challenge for Slurm users is that workload management, provisioning, reporting, and expense management are frequently siloed. By the time users realize they have overspent, they may be deeply in the red, and shutting down spending at this point is akin to closing the barn door after the horse has left.
While Slurm’s Elastic Computing logic and similar third-party solutions can auto-scale cloud resources, Slurm can’t manage what it can’t measure. The scheduler has no visibility to details such as budgets, monthly cloud-spend to date by cost-center, or service costs by cloud-provider. Among Slurm users surveyed, fully 82% of users saw value in associating cloud spending to workloads, but 74% lacked a solution that could automatically associate and manage cloud spending(3).
Cloud automation to the rescue
Fortunately, a new breed of workload-aware cloud automation tools can help automate the complex decisions common to most HPC environments. Cloud automation can help users burst when appropriate, manage cloud spending to budget, and right-size instance selection to minimize cost. Cloud automation works by combining multiple functions into an integrated solution including:
- Cloud service expense management
- Automation via extensible, user-configurable applets
- Multi-cloud resource provisioning and management
- Integrations with leading HPC workload managers
Navops Launch 2.0
Univa® Navops is a multi-cloud provisioning, monitoring, and management platform that helps enterprises integrate HPC environments with the cloud, control spending, right-size cloud resources and automate decisions based on real-time cloud, application and workload-related metrics with composable automation applets.
Unlike simple cloud bursting solutions that provision cloud resources based on static policies or a cloud queue, Navops Launch is application, resource, and budget-aware, and can adjust workload and resource deployments on-the-fly taking into account planned spending by project and department, on-premise capacity, actual cloud resource usage by application, and data access patterns.
With Navops Launch, organizations can maximize and balance the utilization of on-premise and cloud resources, reduce total spending, and boost performance of end-user productivity by ensuring that the right resources are available for the right application.
Navops Launch is production-proven at many of Univa’s enterprise hybrid cloud customers and was recently the subject of a well-publicized million-core cluster deployment on Amazon EC2 by Western Digital.
Whether you are running Slurm or Grid Engine, Navops Launch offers enterprises the ability to deploy and manage hybrid and dedicated HPC clusters for thousands of applications, including AI, machine learning, life sciences, manufacturing, and transportation across multiple clouds while maximizing efficiency and minimizing cloud-spending.
Learn more about Navops Launch and Univa HPC hybrid cloud solutions at http://www.univa.com/products/navops.php.
- Hyperion Research Opinion – March 2019 – https://d1.awsstatic.com/HPC2019/Amazon-HyperionTechSpotlight-190329.FINAL-FINAL.pdf
- Slurm Workload Manager – https://en.wikipedia.org/wiki/Slurm_Workload_Manager – see also, https://www.top500.org/
- Univa Sponsored HPC user survey – pending publication
- Gartner – Ten moves to lower your AWS IaaS costs – https://www.gartner.com/en/documents/3847666/how-to-identify-solutions-for-managing-costs-in-public-c0
- RightScale 2019 State of the Cloud Report from Flexera- https://info.flexerasoftware.com/SLO-WP-State-of-the-Cloud-2019
- Gartner May 24th, 2019 – Are you Ready for Multicloud and Intercloud Data Management? – https://www.gartner.com/en/documents/3923929/are-you-ready-for-multicloud-and-intercloud-data-managem