Make Oracle scream with Flash storage

Eyal MarkovichSenior Director of Performance

A bit about myself

• 20 years APM and database performance experience

• 18 years in Precise working with hundreds of

customers• 4 years in Kaminario managing the performance

team

Performance Analyses Learnings:

1. Business-critical applications require flash 2. Not all flash products are created equal 3. They Require 3D performance consistency

More than 80% of the performance issues are traced

back to poorly performing storage

Let’s start with an example

• Large financial institute• Slow Batch job run for 7 hours• Often exceeding batch window• Critical data not available for the business

20 years: similar disk technology

05/03/2023 03:12:29 PM

6

Databas e

Selecting 1% of the database …

1990

1 Min

1.4 GB

1994

2Min

4 GB

1998

3.5Min

9 GB

2003

18Min

73 GB

2011

11.5 Hours

2 TB

Data:1400X

Performance:5X

Storage Architecture Evolution

- Real Time, elastic and agile

- 3D performance

- Resilient

- Scalable

- Best TCO

- Under utilized capacity

- No shared storage

- Difficult to grow

- Creates silos

- Hard to manage

- Same as DAS

- Tuned for single workloads

- IOPS focused only

- Inconsistent

- Not resilient

- Not scalable

- Over provisions $$$- No shared workloads

- Can’t scale easily

-Disrupted Performance Limited

-Doesn’t fit flash

-Performance Limited

XIV

Flash – Competitive Landscape3D Performance

SoftwareCapabilities

Silo’d ($$$ & Resources)One WorkloadNo SW stackNo Scalability No enterprise HA

Silo’d ($$$ & Resources)Limited WorkloadsLimited SW stack or 3rd partyLimited ScalabilityNo enterprise HA

Silo’d ($$$ & Resources)Limited WorkloadsLimited ScalabilitySLA under failure

Mixed WorkloadsFull SW stackEasiest management Best TCOEnterprise HA

Back to the example

• Large financial institute• Slow Batch job run for 7 hours• Often exceed batch window• Critical data not available for the business

I/O Wait beforeKaminario

I/O WaitAfter Kaminario

CPU

What about my Oracle instance?

• Do I need Flash for my Oracle?

• What type of Flash is best for me?• SSD cards? Hybrid array?• SAN flash array? Scale out SSD flash array?

8:00 am

CPU I/O WaitsLock Net

9:00 am

What about my Oracle instance?

• Do I need Flash for my Oracle?

What will be the effect of Flash on my workload?

What will be the effect of Flash on my workload?

db file sequential read

What will be the effect of Flash on my workload?

db file scattered read

What will be the effect of Flash on my workload?

Direct Path Read

Sample SQL Query:Select /*+ Parallel(emp 4) */ * from Employee emp;

Execution Plan:SELECT STATEMENTPX COORDINATORPX RECEIVEPX SEND RANGEPX BLOCK ITERATORTABLE ACCESS FULL EMPLOYEE

What will be the effect of Flash on my workload?

Log file sync wait

What will be the effect of Flash on my workload?

Direct path read temp and direct path write temp

Performance Analysis Issues:

• Random read latency for small block size• Large latency for sequential reads• Mixed workload applications running both OLTP & reports• Parallelism set too high or too low• Writes (DML) affecting read latencies• Storage unable to support temporary workload• Large latency for writes

Inconsistent Performance Across Workloads

SAP Example: Simple Workload

Running Overnight• Large sequential scans• High parallelism on the database• High throughput

Two Distinct Workloads:1. Dialog User activity2. Programs (batch)

SAP Example: Simple Workload

What we need from storage to improve the performance:

Throughput1,483 MB/s

IOPS4,410

Latency0.89 ms

HIGH

LOW

SAP Example: Complex Workload

Can you run the SAP programs during the day?Activity:

• What will the effect be on User activity?• How well will the SAP programs perform?

It’s All About Storage:• What will the effect be on storage latencies?• What will the effect be on storage throughput?• Will storage performance remain consistent ?

Can your system tolerate mixed workload?

SAP Example: Complex Workload

You need flash that supports mixed workloads:• Consistently low latencies regardless of the throughput• Delivering high throughput regardless of the IOPS

HIGH

LOW

Throughput1,713 MB/s

IOPS24,426

Latency.56 ms

Storage done better

Scale-Out

Granular & Linear (6- 120 TB)

Automatic performance & endurance

7 year flash endurance Simple & Intuitive

Global management Low TCO than

alternatives

What are Kaminario K2 differentiators?

No Single Point of Failure

Self healing – N+1 Instantaneous, space

efficient, snapshots Non disruptive

upgrades Full node recovery

~ 90 minutes

Safe & Reliable

Consistent Lowest Latency (280 us) Linear scale (200K- 2M+

IOPS, 3-30 GB/s) Mixed workload

concurrently Minimal performance

degradation (<25%) during recovery

3D Performance

Kaminario K2 Building Block

Enterprise Grade Highly Available Building BlockCluster together to scale-out Performance or Capacity

Feature Specification

Flash 8 MLC Flash (800 GB or 1.6 TB) per K-Node

Usable Capacity 3 TB / 6 TB per K-Node(6 – 12 TB) or (12- 24 TB) per K-Block

Performance IOPS - 200K-410K Throughput - 3–6GB/sLatency - 280 us

Footprint 6 RU – 8 RU

Power Typical 1.4kW-2kW

High Availability System components fully redundantDual Processors, dual battery

Host Ports 2 host ports per K-Node8 x 8Gbp FC HBA or 8 x 10GbE iSCSI per K-Block

K-BlockK-NodeK-NodeK-NodeK-Node

Customers Span Industries & Use Cases

SPEAR Technology - Scalability

Flash to the Rescue

Commodity HW. It’s all about SW architecture

SPC-1 World Record Performance

(.80 cents per SPC-1 IOPS)

SPC-1 2013 Configuration

CONFIGURATION

• 86 TB System• 7 K-Blocks• 34U Footprint• SMART – All Flash

CONFIGURATION RESULTS• 3 Million Read IOPS at under 2ms latency

• 2.8 Million Read IOPS at under 1ms• 1.2 Million Write IOPS at under 1ms

• 56 GB/s Read/Write Throughput

3D Performance: Latency under failureKaminario Scale-Out Flash ArrayHybrid & Monolithic Flash

Get a Performance Assessmentwww.kaminario.com/performance