FlashGrid: Database Infrastructure Simplified

Simple, fast, and highly-available database infrastructure for public cloud, private cloud, and physical environments.


The 42nd IT Press Tour had the opportunity to meet with Art Danielov, CEO of FlashGrid. Their vision is to be the trusted provider of database infrastructure solutions on modern cloud and hardware platforms.


The Oracle database is still the number one database in use. It has a dominant position in high-end OLTP and is heavily used in financial services, government, and telecommunications.


Moving Databases to the Cloud


FlashGrid began as software-defined storage for Oracle DB on standard servers but customers wanted the same in the cloud. High availability and high-performance challenge mission-critical transactional databases in the cloud.


DBaaS offerings are good but often insufficient due to limited high availability, limited performance, limited control, compatibility issues with legacy apps, and cloud lock-in.


The FlashGrid Cluster


The FlashGrid cluster is an engineered cloud system for running the Oracle database. It provides high-availability clustering (active-active and active-passive), extreme performance options on AWS, Azure, and GCP. It offers automated infrastructure-as-code deployment and has multiple customers in finance, payments, billing, government, eCommerce, and healthcare in production since 2017.


The FlashGrid Cloud Area Network creates a high-speed network overlay with multicast, VIPs, and QoS. FlashGrid Storage Fabric creates shared storage from locally attached disks (EBS or local SSD). Proven Oracle Clusterware and ASM for high availability and data mirroring.


Extreme Performance Options


FlashGrid offers the following:

  • 22.5 GB/s storage throughput on R5b instances

  • 48+ GB/s of storage throughput with local SSDs on i3en.metal instances

  • 100 GB/s network with r5n, i3en, and other instance types

  • 224 physical cores and 24TB on high-memory instances

  • Multiple database and storage nodes for extreme throughput and high capacity

Extra-Large Oracle RAC DB on AWS:

  • 10 GB/s per DB node

  • Fast local SSD capacity: 60 TB raw per storage node (before mirroring), 100 TB usable with 5 storage nodes, 300 TB usable with 15 storage nodes

  • Unlimited EBS capacity

  • Any number of DB and storage nodes

  • DB nodes: r5n or High Memory (100 Gb/s network)

  • Storage nodes: i3en, m6i, or c5

  • Each storage node in a separate failgroup

  • Can tolerate the loss of any two storage nodes

Extra-Large Oracle RAC DB on Azure:

  • 8000 read + 8000 write MBPS per DB node (E104is_v5)

  • Up to 4000 MBPS per storage node (D96ds_v5)

  • Any number of DB and storage nodes

  • Up to 128 TB (32 x 4TB) raw per storage node

  • Each storage node in a separate failgroup

  • Normal Redundancy: 2x data mirroring, usable capacity = 1/2 raw capacity, Can survive the loss of one storage node

  • High Redundancy: 3x data mirroring, usable capacity = 1/3 raw capacity, Can survive the loss of two storage nodes

The FlashGrid launcher enables a fully-functional cluster in 90 minutes with infrastructure-as-code deployment where you configure once and deploy multiple clusters thereby eliminating human errors in high-availability configuration.