SPEED PowerMax Exam Flashcards

1
Q

Which type of write operation is most efficient for the system and requires the least amount of system resources to perform?

A

write on a track that is already marked as write pending

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Using the output of the symcfg list - v command below, how many slots are available for caching write operations?

A

130.2

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

How does a Delayed Fast Write (Sustained Write) differ from a Fast Write (Max Write)?

A

The fast writes are the maximum write rate measured, usually when the write pending count is very low. The delayed fast write is the rate we can sustain writes for a long time, usually at the write pending limit

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Choose the correct statement about mainframe workloads as they compare to open system workloads

A

mainframe systems typically have a higher percentage of read hits and often require more front end resources than back end resources

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

The architecture of the PowerMax pools CPU resources. How are these resources allocated to specific emulations?

A

when the system is initialized CPU cores are allocated based on the different emulation types that are configured and other configuration considerations

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

on a PowerMax, a host is writing a 64KB block and the Logical Block Address is not aligned with the Cache Slot Size. What is the impact to system performance?

A

it may increase both cache utilization and IO to physical disk, and this could increase resource utilization and potentially impact system performance

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

How do you minimize the chance of misaligned IO by a host connected to a PowerMax?

A

using Host software, align the partition so that data begins at an offset in multiples of 128K

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Under which of the following conditions is a PowerMax system said to be unbalanced?

A

when the average utilization of some system components is siginificantly different from that of components of the same type

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

On a PowerMax, how many disk operations are involved when a backend director performs an uncompressed 16K write on a RAID5 4+1 data device?

A

there would be 4 disk IO operations: two reads and two writes

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Which statement is most accurate about how RAID protection is configured on a PowerMax system?

A

physical disks are logically divided into hypers and hypers are used to create RAID protected data devices (TDATs)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Which RAID protection has the least overhead for small block read operations and would provide the best system performance given the same number, type, and size of the physical disks?

A

All three have roughly the same small block read performacne

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

If a workload is spread evenly across all logical block addresses of all devices, what is the skew?

A

No workload skew

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What is the benefit does RAID protection provide in both VMAX and PowerMax?

A

eliminates the disk drive as a single point of failure

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What is the primary advantage of RAID-6?

A

Availability

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

A workload has a skew of 85/20. To what is this referring?

A

85% of the IO activity is being performed against 20% of the capacity

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Which workload would be most efficient for RAID-5?

A

large block sequential writes

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

There are two choices for configuring RAID 5: 4+1 and 8+1. For the same number and size of drives, which configuration would provide the better performance for a random small block read and write application?

A

during normal operations there is no difference in performance

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

There are two choices for configuring RAID 6: 4+2 and 14+2. For the same number and size of drives, which configuration would provide the better performance for a random small block read and write application?

A

RAID 6 6+2 would be better as there would be more drives

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

You have been asked to size a PowerMax system and were told that the system needs to be able to do at least 180K IOPS and present a minimum of 300TB of storage. What other information do you need to start sizing the system?

A

the workload characteristics

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

Which object is a logical grouping of TDEVs?

A

storage groups

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

What is the best option for designing a PowerMax system for maximum availability?

A

mirror the production PowerMax in an SRDF/S configuration

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

Which statement is true about the machine learning on PowerMax?

A

designed to position data to minimize impact of data reduction based on historical statistical information

23
Q

Which user task is required when configuring the logical storage on a PowerMax?

A

create TDEVs and Storage Groups, add TDEVs to group, and assign SLO

24
Q

A customer is observing an average of 8800 IOPS on each of four frontend ports of a PowerMax and is asking if this is an acceptable number and if they are approaching the limit for a port. What would be the best response?

A

Explain the IOs per second is only one aspect of a workload; further investigation is required

25
Q

An application normally does around 11K IOs per second but burst up to 48K+ IOs per second. During the bursts, it has a negative effect on the response time of other time sensitive applications that share the same four frontend ports. What is the best way to regulate host IO arrivals to minimize the impact on the system during these bursts?

A

Configure Host IO limits on the storage group used by the bursty and less important application

26
Q

Which statement correctly defines the impact of frontend queuing on a PowerMax system?

A

some queuing may be acceptable but may increase response time

27
Q

To achieve the best possible performance and availability on the frontend of a four node PowerMax system, what would be the host choice of Fibre Channel Directors and ports to use to connect a host?

A
28
Q

The customer is planning a new PowerMax system and insists on knowing the number of IO per second that an OR can perform. What would be the best way to respond?

A

Ask for clarification on the workload characteristics and reference SPEED materials that report results for specific workloads

29
Q

What happens when a PowerMax system reaches the System Write Pending (WP) Ceiling?

A

host writes will be delated except writes to tracks that are currently marked as write pending

30
Q

What is the first step of a performance investigation?

A

get a clear problem statement

31
Q

When the PowerMax is at the System Write Pending Ceiling, what is the expected impact?

A

all new writes will experience a significant increase in response time

32
Q

Why does more granular performance data that is collected using shorter intervals help in performance analysis?

A

will display peaks of activity that are lost in averaging

33
Q

While the performance of the system is very good, the response time for an application has been inconsistent and you want to determine if Write Pending Ceilings could potentially be a problem. Which metric in Unisphere for PowerMax Performance charts would you use?

A

system/device WP events/sec

34
Q

A set of servers is driving 768MB and 24K IOs per second. What is the average block size?

A

32KB

35
Q

if you had two different workloads, one being completely random and the other sequential, how would you expect these workloads to perform on a PowerMax system?

A

sequential workloads will get more cache hits and will likely have lower response time

36
Q

Which statement correctly describes write pending limit events on a PowerMax?

A

A PowerMax may see device write pending events even though the total write pending count is low

37
Q

A workload has a tight locality of reference. What is the effect on system activity?

A

the system is likely to get more cache hits

38
Q

A set of servers connected to four 32GB OR frontend ports is driving a total of 24K IOs per second w/ an average block size of 32KB. What would you expect the throughput to be for each of the frontend ports?

A

188MBs per port

39
Q

Which workload type would result in the lowest cache hit rate reported by Unisphere?

A

random read

40
Q

A workload is said to have a high locality of reference. What does this mean?

A

IOs are clustered around a small range of logical block addresses

41
Q

How is a write hit defined in Unisphere on PowerMax?

A

write to cache without any delay

42
Q

A PowerMax was sized for 50K OLTP IOs per second and the customer’s tests show the system is achieving considerably less. What could be the reason?

A

The customer’s tests are not representative of the sized OLTP workload

43
Q

If a single threaded application has a 5ms response time what is the maximum IO rate?

A

200

44
Q

If a single threaded application has a 1ms response time what is the maximum IO rate?

A

1000

45
Q

The example of a DB simulation of a PowerMax shows 8K random read hit. What is the maximum IOPS on a 4 node pair PowerMax?

A

5,227,048

46
Q

Which represents a DSS type workload?

A

Workload C

47
Q

What is a characteristic of an OLTP type application?

A

workload is mostly random small block IO

48
Q

How does the Performance Engineering Group define a write hit in their testing results?

A

only writes on write pending tracks are considered hits

49
Q

The example of a DB Simulation Chart for a PowerMax shows the results of OLTP testing for node pair counts. What do the vertical solid lines indicate?

A

maximum IOs per second sustainable

50
Q

Your objective is to measure the capability of a frontend port. Which application type would generate the least amount of IO and thus be inappropriate for a measurement?

A

a single application thread that waits for the IO to complete before sending the next

51
Q

You are a System Architect working with a customer who asked for your help with a critical performance issue on a production box. What action should you take to engage the support process?

A

Get a clear problem statement and ensure the customer opens an SR

52
Q

Which statement concerning the collection of .dcf performance data on a PowerMax system is correct?

A

data collection on a POwerMax runs on the Guest OS

53
Q

When collecting data for the purpose of performing a performance analysis, what data do you need?

A

a data collection that includes the period that covers when the problem occured

54
Q

Which configuration file describes the hardware configuration of the system adn the logical configuration, including devices, pools, SRP, and GOS?

A

The IMPL.bin file from the system that is downloaded to the Control Station