A customer asks an architect to help troubleshoot low throughput in their high-performance compute (HPC) environment. All 70 HPC nodes have a single 10Gb connection to a 96 port 10Gb switch. The FlashBlade is connected to their dedicated HPC switch with 8x10Gb connections. The HPC application is using a single shared S3 bucket for the data being processed.
Which change is needed to increase throughput?
A customer is adding a micro-service to an existing application and needs to use the same user and S3 bucket. The current secret key is unknown.
What should the customer do to resolve this issue?
A customer is implementing a new backup system using four backup appliances, with dual 10Gbps NICs writing to a FlashBlade. Each backup appliance can write at line speed.
What is the minimum number of 52TB blades to be able to handle this workload?
A customer wants to use FlashBlade as storage for a business critical, high-traffic SQL server. Why will this architecture fail?
A developer of a business-critical application is concerned about system reliability when it comes to a IO resistance. An architect needs to explain the FlashBlade write mechanism and address three main concerns:
- Data location when FlashBlade acknowledge a write
- Number of created data copies before FlashBlade acknowledges a write
- Blade failure resilience
When does write acknowledgement happen?
