ONTAP Discussions
ONTAP Discussions
Does anyone have a whitepaper or collateral material that addresses competitor FUD on NetApp performance? Specifically undersized CPU?
I'm a partner in pre-sales, and I have a customer that is keyed-in to a competitor's FUD regarding NetApp performance in general. The competitor (don't know source) is claiming that all NetApp systems are underpowered on CPU side, and when you add disk the system gets worse. I've had discussions with the customer trying to refocus them on what's important in disk subsystem performance design. First of all, the system they are planning to purchase is all SATA. I've made other points about number of disks, full-width write design of ONTAP/WAFL. I've also attempted to get more info on their application and requirements...stating that we will assure them they have the right design for their usage. E.g., small random reads/writes or large sequential reads/writes, read-to-write ratio, number of IOPS, throughput requirements, etc. Still no success.
I have performance info that is not customer-ready (proprietary). Although, most of this material is not useful to combat this customer's concerns (whether valid or not).
Your help sincerely appreciated.
Solved! See The Solution
You're being asked to prove a negative, which is a tough proposition. If you can't turn the assumption around ("Make them prove that the CPU is underpowered"), then you're only bet may be to get some references and let your customer talk to other customers in a similar industry or a similar configuration/application. I think some customers look at CPU because they think it's easy. In reality, CPU is rarely the true bottleneck and a high CPU utlilization doesn't necessarily mean you have a problem since ONTAP will use free cycles to run background jobs that will back off if real work comes in. I only say this because if you do win this, you want to disuade them from constantly monitoring CPU as a measure of capacity or performance. Bottom line, you need to look at load in terms of network in/out and response time to clients.
It's like a router. You probably don't care what CPU is in your router, rather you care that it can route X amount of traffic.
But bottom line, if they don't believe you, maybe they will believe another customer.
You're being asked to prove a negative, which is a tough proposition. If you can't turn the assumption around ("Make them prove that the CPU is underpowered"), then you're only bet may be to get some references and let your customer talk to other customers in a similar industry or a similar configuration/application. I think some customers look at CPU because they think it's easy. In reality, CPU is rarely the true bottleneck and a high CPU utlilization doesn't necessarily mean you have a problem since ONTAP will use free cycles to run background jobs that will back off if real work comes in. I only say this because if you do win this, you want to disuade them from constantly monitoring CPU as a measure of capacity or performance. Bottom line, you need to look at load in terms of network in/out and response time to clients.
It's like a router. You probably don't care what CPU is in your router, rather you care that it can route X amount of traffic.
But bottom line, if they don't believe you, maybe they will believe another customer.
Adam has completly right!
Don't argue with your customer. Don't try to convince him. Let him talk with an other customer and your problem is solved is a few seconds.
Greetings, Reinoud
Reinoud, are you informally asking to be a reference?
Christopher
Only when the customer is located on a tropical island, with a private beach and a cool bar.
I am a customer and used to work on EMC and HP system's and have learnt to take everything I am told with a pinch of salt. Good luck trying sort out the FUD. I recommend beer and a white board.