Performance metrics measured by one of our customers running H2O.
Performance Is Paramount
Bigstep Metal Cloud is able to push performance further than any virtualized cloud. Our infrastructure uses bare metal servers, fully isolated wire-speed networks, and high-performance storage systems to deliver unparalleled benefits in terms of performance, price, and security.
Bigstep Metal Cloud is architected to champion data-intensive applications and deliver all the power you need when you need it.
Easy to Get Started
Use the pre-defined configurations below to get started
or log in to fully customize your infrastructure.
Bigstep Compute Instances
Bigstep Servers are fine-tuned right down to the metal. We handpick the best hardware components, make specific firmware choices, and configure BIOS settings to get the most out of each chip on that motherboard. Given that the OS resides on a distinct Compute Instance, storage is quickly scalable to eliminate cost-per-use or to update configurations without loss of data. If needed, they can be restarted within minutes. The Bigstep portfolio includes NVMe local storage servers, physical drives, and GPUs.
- Pre-configured for maximum performance
- Single-tenant, bare metal servers
- Full control: root access, out of band and OS
- One-click upgrading and scaling
Bigstep uses a distributed Block Storage architecture, with several small-sized, independent SANs, directly connected to the racks for the lowest latency, ensuring isolation of any incident. To achieve the right mix of performance and capacity, we provide two tiers of block storage:
Solid State Storage (SSD)
For high-performance workloads - each
drive delivering up to 500K IOPS/sec
Disk Storage (HDD)
For cost-effectively storing massive data
sets - up to 10 Gb/s throughput per drive.
- Large scale HTTP file system
- Cost-effective storage
- Fully HDFS compatible
- Supports files of any size
- High throughput and low latency
The Bigstep Network is unmatched in performance and flexibility by any other cloud network. Cut-through switching ensures wire-speed latency and complete control over the physical ports in each switch. It's even safer and much more elastic than on-premises deployments.
- The Bigstep Network is the world’s highest performance public cloud network
- Wire-speed transfer achieved with cut-through switching
- Isolated software-defined physical LANs created with one click
- Unlimited Layer 2 broadcast domains for each user
- Instance ports aggregation, providing up to 40 Gb/s to a single destination
Bare Metal Containers as a Service
Bigstep Container Service is a Docker-based service that enables users to multiplex single-tenant bare metal machines and run hundreds of independent workloads and applications on the same physical hosts. Moving to a Docker-based infrastructure does not entail rewriting your applications for microservices. Any Linux-based application can be converted to run on the container platform, instantly gaining resilience, performance, and automatic load-balancing out-of-the-box.
Frequently Asked Questions
But my app runs just fine on AWS. Why should I move to Bigstep?
Apart from typically halving the costs, we also make it so much easier to work with integrated technologies. We help you scale when you need to, by adding extra nodes and configuring them to work as a cluster.
Does Bigstep offer managed services?
Yes, we offer what we call the “Single Point of Contact” support plan. We will work with the software vendors and solve all your issues. This saves you from the back and forth between the infrastructure vendor and the software vendor when performance issues arise.
How do I connect securely to Bigstep Metal Cloud?
We offer a VPN service that connects to your current infrastructure (such as AWS or on-premises data centers). Bigstep Metal Cloud supports standard ipsec protocols and is fully compatible with most enterprise security standards.
What is the latency between Bigstep and my AWS region?
Typically, we are around 5 to 10ms away.
What’s the secret sauce behind Bigstep's 2-5x performance improvement?
Latency matters a lot in analytics. In virtual clouds, it creeps in, induced by the hypervisor, by the many layers of the virtual network, storage caching techniques, etc. We have removed every element that could introduce latency by using only bare-metal nodes, 4x10Gbps cut-through east-west optimized network, PCI-e attached NVMe storage and distributed SSD-based block storage. We are squeezing every bit of performance out of the hardware without sacrificing flexibility.
Will low latency increase my app’s performance?
Not all workloads require a low-latency setup. It all depends on the latency tolerance of your application.