VAST Data

VAST Data has developed a new type of all-flash data storage (Universal Storage) that is ideal for HPC and AI applications or fast data recovery after ransomware attacks.

VAST Universal Storage

VAST Data has launched a versatile data storage solution based on flash technology. The combination of SCM (Storage Class Memory) and QLC memories makes it possible to achieve the high performance of an all-flash system at a price comparable to disk arrays with spinning hard disks (HDD). The disaggregated, shared-everything architecture offers high reliability and unparalleled scalability. A complete Software defined Storage (SdS) approach allows you to add new features and support the latest storage technologies and protocols. As one of the first manufacturers on the market, VAST Data offers online compression and deduplication for efficient data storage, support for Nvidia GPUDirect Storage (GDS) technology for direct data transfers between GPU and storage, or integration with Veeam and Comvault backup software. VAST Data is currently one of the fastest growing storage vendors on the market and M Computers represents it in the CEE region.

Flash for price of HDD

The storage is built on QLC technology (SSD). Together with the compression algorithms, the price is close to hard disk drives (HDD).

Scalability

The VAST architecture provides almost unlimited capacity and performance scalability by adding additional blocks of storage.

10 years lifetime

VAST guarantees support price for 10 years and interoperability between older and new technologies within one storage.

Native support of S3

Implementation of object storage is just easy on VAST Data solution thanks to the native support of the S3 protocol.

VAST architecture

Disaggregated, Shared-Everything architecture

VAST Data storage is designed as a so-called. Disaggregated, Shared-Everything architecture (DASE) Current system state (metadata) and storage capacity (data) are built on high-density NVMe-oF (NVMe over Fabric) all-flash disk shelves (DNodes). All system logic and data management is implemented as state-less Docker containers that run on computing servers separately from the data layer (CNodes). The term disaggregated refers to the independence of the data layer from the compute and their independent scaling of performance as needed – adding disk or compute polices. Shared-everything refers to the fact that each computing server (container) has access to the entire data space (both data and metadata). This eliminates (east-west) communication between compute servers (controllers), which improves performance and scalability (up to thousands of servers in a cluster).

100% non-volatile storage

VAST storage is designed entirely on non-volatile technologies – super fast SCM memory for caching and metadata and QLC chips for data storage. In the event of a power failure, no data will be lost. SCM (Storage Class Memory) is a flash memory similar to conventional DRAM with low latency and high transfer rate.
However, unlike DRAM, SCM storage is non-volatile, i.e. it does not lose data when power is lost. QLC Flash storage is the most economically viable flash memory. In combination with SCM, excellent performance parameters can be achieved while maintaining the high lifetime of QLC chips by efficiently writing data from SCM to QLC. VAST recently announced multi-vendor support for SCM and QLC memory.

Outperforms even traditional disk solutions with price

The VAST solution is based on the current best technologies in the field of data storage — a combination of SCM and QLC memories and, above all, a complete Software Defined Storage (SdS) approach. Together with its unique data compression algorithms, it provides the best TCO solution on the market that can outperform even traditional spinning hard disk arrays. To give you an idea of the cost savings, VAST has developed a data TCO calculator that can show the savings of deploying a VAST universal disk array.

VAST TCO Calculator

99.99999% solution reliability

A reliable and safe solution

Another significant advantage of the VAST solution is its reliability. The system is designed in such a way that the failure of any component does not represent a failure of availability of the entire system. In addition, the failure of compute nodes (CNode) only affects the speed but not the availability of the service.

Data replication

For deployments in Enterprise environments, VAST Data offers reliable data replication to other sites or to cloud storage through a fast and secure S3 interface. All of these backups do not take up as much space or require high bandwidth requirements due to the implementation of advanced compression and deduplication technologies.

Full hardware redundancy

Has there been a power failure? Do you have full disks? Did all your servers crash? VAST’s solutions are ready for all these cases and will keep your system running at full capacity under any circumstances. Redundancy and mechanisms to recover from the failure of a part of the system are one of the main advantages of VAST solutions. This is also true when updating the entire system, where there is no need to schedule downtime as VAST is equipped with a Non-disruptive updates (NDU) update system. The architecture is designed in such a way that the individual components are updated gradually without affecting the availability of the system.

Virtual IP addresses and erasure codes

To ensure high availability of the VAST solution is to implement technologies to spread the load across all available resources. This is taken care of by the Virtual IP Addresses (VIPs) service, which assigns a range of addresses to access servers (CNODEs) through which clients or servers then connect to data. System administration does not have to deal with translations and changes of physical addresses in case of server failure. In addition, the data flow is optimized to distribute the load evenly over the available resources. Erasure encryption, combined with other technologies, ensures high data resiliency in the event of a failure of the physical flash drives in the VAST Data repository. Since the data being recovered is already reduced, the recovery speed is significantly faster than traditional disk arrays.

The best data reduction on the market

Increased lifetime of QLC chips due to local block compression

Flash storage can overcome the liveliness of traditional hard drives, but only under certain conditions. One of the ills of traditional solutions is the alignment of data into “blocks”, where the block size for storing data is fixed. In normal operation, these blocks fill unevenly and highly inefficiently, causing premature wear and later eventual malfunction. VAST Data solves this problem by Adaptive Chunking of the data, i.e. splitting the data to fill the maximum cell capacity. By implementing variable block sizes, it creates blocks of data of the ideal size for specific (and large) data. The overall size of the blocks themselves is further reduced by compressing them on local storage — the block size can even adjust to the size of similar blocks stored on the datastore, further optimizing data compression.

Global block deduplication

Similarity reduction

The VAST data repository not only has unique online compression and deduplication mechanisms, but also offers compression algorithms based on data similarity. Instead of the usual compression, it can split the data into parts of different lengths and compare them with each other. If it finds a repeating pattern among these data, it simply keeps that part of the block once and references the remaining instances to that data.

VAST Probe

Wondering how much space you could save with VAST’s unique compression and deduplication technologies? VAST data offers a simple tool called VAST Probe, which, in the form of a Docker container, accurately analyzes your data and determines the realistic compression ratios that can be achieved.

Installation of VAST Probe

VAST Data System Administration

Versatile and simple control panel

With the ability to manage your VAST system through the app via Web UI or API, you can monitor the status of your system from anywhere. The WebUI Dashboard can be divided into 4 basic tabs, under which you can find the most important information about the system: Capacity, Performance, Inventory or Graphs for system metrics.

1

2

3

4

5

6

7

Competent documentation and manual for Web UI operation here.

VAST Gemini

VAST storage can be purchased in the Gemini model. Certified hardware is supplied directly from the manufacturer at cost. The software part takes the form of a subscription for the required capacity and number of years. The customer thus buys an appliance for which VAST data covers both hardware and software support for up to 10 years from the date of purchase. VAST thus guarantees software compatibility with new and older versions of computing (CNode) and data (DNode) units.

Machine Learning

VAST Data collaborated with NVIDIA to develop the system. Together they created the VAST DGX POD reference architecture for the most demanding applications in the field of artificial intelligence. NVIDIA adds to this architecture its DGX systems (AI appliance) and networking (Mellanox), VAST data scalable all-flash storage. VAST Data was among the first in the market to offer support for NVIDIA GPUDirect Storage(GDS) technology for fast data transfers between GPU accelerator memory and data storage.

Data recovery

In the age of ransomware attacks and the threat of large volumes of data being encrypted, fast recovery is essential. Modern data backup scenarios assume a first backup layer on flash drives. VAST data offers integration with data backup and recovery tools (Veeam, Comvault and others). Thanks to its architecture, there is no problem to recover data at speeds of tens of GB/s. Compression algorithms and high reliability offer the best TCO for deployment in Enterprise environments of banks, industrial enterprises, or government.

Webinar: Inside VAST storage (in Czech)

We have prepared a detailed description of the VAST data solution and a webinar to introduce the unique features of VAST storage.

Presentations to download

Our Certified VASTronauts

1 2 3 4

Petr Plodík

M: 737 264 480 petr.plodik@mcomputers.cz

VAST Data

VAST Universal Storage

Flash for price of HDD

Scalability

10 years lifetime

Native support of S3

VAST architecture

Disaggregated, Shared-Everything architecture

100% non-volatile storage

Outperforms even traditional disk solutions with price

99.99999% solution reliability

A reliable and safe solution

Data replication

Full hardware redundancy

Virtual IP addresses and erasure codes

The best data reduction on the market

Increased lifetime of QLC chips due to local block compression

Global block deduplication

Similarity reduction

VAST Probe

VAST Data System Administration

Versatile and simple control panel

VAST Gemini

Machine Learning

Data recovery

Webinar: Inside VAST storage (in Czech)

Our Certified VASTronauts

Petr Plodík

News

Another edition of the AI DAYS technology conference took place

AMD EPYC™ servers for brain dynamics research

NVIDIA DGX at CEITEC BUT in Brno

Weekend with the Heart on the Palm Foundation

About us

NVIDIA products

Intel and AMD products

Contact