A recent Information Week’s 2014 State of Enterprise Storage Survey indicates that 40% of respondents were using SSDs in disk arrays and about 39% deployed SSDs in servers, which is about 8-10 points up from the previous year. These figures clearly highlight that SSD adoption is increasing.
We brought together a group of expert DBAs, geeks and IT/System administrators to share some tips on configuring SSDs and storage for enhanced SQL Server performance. They joined hosts SQL Server MVP and SolarWinds Head Geek Thomas LaRock (@SQLRockstar) and Joseph D’Antoni, (@jdanton), SQL Server MVP and Senior Architect at Denny Cherry & Associates. We posed some key questions to the participants on the topic and here are some key takeaways from the conversation (you can read the full chat transcript at the end of this article).
Tip: Cost is the number one factor driving increased SSD adoption rates
Datachat participants were unanimous in stating that reduced cost was the primary reason driving increased adoption rates for SSDs. Increased demand for virtualization has invariably led to increase in demand for SSDs. The other reasons for adopting SSDs are: lower mean failure rate than spinning media and lack of developers to tune virtualized environments.
@SQLRockstar @solarwinds It’s interesting–based on data I’ve seen, SSDs have a lower mean failure rate than spinning media 1/2 #datachat
— Joey D’Antoni (@jdanton) March 10, 2015
Tip: It’s a myth that SSDs solve all performance problems
Because of the decreasing costs for SSD storage, it’s not uncommon to attempt to solve performance issues by adding SSD storage. But, as @SQLRockStar lamented, “Disk is rarely the problem: Its often bad code/design, or bad disk configuration.” New hardware such as SSDs drives only solve problems caused by disk. Joseph D Antoni also noted:
.@solarwinds A4: If everything is taking a lot of money out of you account, yes. They are really good at a lot of things. 1/2 #datachat — Joey D’Antoni (@jdanton) March 10, 2015
But, most agreed that SSDs do not offer a solution for many common problems such as:
- SAN problems
- Bad application
- Hypervisor paging
- Bad code and design
- PEBKAC issue
Tip: CPU over allocation is a key performance metric for troubleshooting VMware
If you’re trying to troubleshoot performance of SQL Server on VMware, the key metric is the number of CPUs. Over-allocation of CPUs is bad. For SQL Server, it’s fair to look at all the same metrics that you would on physical servers. If you do not get your answers from this analysis, then it’s time to analyze the host. Some other metrics that help with troubleshooting include:
- VM CPU ready time
- SQL Server wait events
- Activities in vCenter/SCVMM
Tip: Common challenges with SSDs: include index rebuilds and sequential writes
Common complaints with SSDs are that they aren’t good at index rebuilds and sequential writes. These are some unique challenges that are faced by tiered SANs. An improper DR (disaster recovery) plan can only add to your woes.
Full datachat transcript
| solarwinds | Q1: SSDs are not new tech, but becoming more popular. What factors are driving adoption rates? #datachat |
| SQLRockstar | RT @solarwinds: Q1: SSDs are not new tech, but becoming more popular. What factors are driving adoption rates? #datachat |
| jdanton | @solarwinds A1: Cost have gotten a lot cheaper. I was looking at an old deck that had some pricing. And it’s huge. #datachat |
| SQLRockstar | .@solarwinds A1: Costs, mostly. But there is also the belief that that hardware is more durable than it was previously. #datachat |
| jdanton | @solarwinds A1: Also, virtualization tends to increase demands on a storage subsystem, and SSDs can help with that. #datachat |
| jdanton | @SQLRockstar @solarwinds It’s interesting–based on data I’ve seen, SSDs have a lower mean failure rate than spinning media 1/2 #datachat |
| jdanton | @SQLRockstar @solarwinds but perception is still there that it’s an issue. #datachat |
| DBABullDog | @solarwinds @SQLRockstar A1 IMOO cost would be number 1 factor. Lack of developers to tune environments could be another factor #datachat |
| SQLRockstar | .@solarwinds A1: Also, at this point, there is a lot of material on best practices for SSD usage and deployment. #datachat |
| DBABullDog | RT @jdanton: @SQLRockstar @solarwinds It’s interesting–based on data I’ve seen, SSDs have a lower mean failure rate than spinning media 1/… |
| SQLRockstar | RT @DBABullDog: @solarwinds @SQLRockstar A1 IMOO cost would be number 1 factor. Lack of developers to tune environments could be another fa… |
| solarwinds | RT @jdanton: @SQLRockstar It’s interesting–based on data I’ve seen, SSDs have a lower mean failure rate than spinning media 1/2 #datachat |
| SQLRockstar | .@jdanton Yeah, the perception of high failure is still there, I think. #datachat |
| SQLRockstar | RT @jdanton: @SQLRockstar @solarwinds It’s interesting–based on data I’ve seen, SSDs have a lower mean failure rate than spinning media 1/… |
| solarwinds | @DBABullDog Welcome back Daniel! #datachat |
| solarwinds | Q2: What’s the number one mythunderstanding about SSDs? #datachat |
| jdanton | @SQLRockstar The vendors have done a really good job writing firmware to do write balancing. #datachat |
| SQLRockstar | RT @solarwinds: Q2: What’s the number one mythunderstanding about SSDs? #datachat |
| jdanton | @solarwinds A2: 1) They fail frequently 2) They have to cost a ton of money 3) They cure all ills. #datachat |
| SQLRockstar | .@solarwinds A2: That SSDs will solve all your performance problems. #datachat |
| DBABullDog | @solarwinds glad to be here #datachat |
| jdanton | @SQLRockstar @solarwinds But the SAN vendor sold me these magic beans with my SSDs. #datachat |
| DBABullDog | RT @SQLRockstar: .@solarwinds A2: That SSDs will solve all your performance problems. #datachat |
| SQLRockstar | .@solarwinds I rarely see disk as the bottleneck, it’s often bad code/design, or bad disk configurations. #datachat |
| solarwinds | RT @SQLRockstar: .@solarwinds I rarely see disk as the bottleneck, it’s often bad code/design, or bad disk configurations. #datachat |
| SQLRockstar | .@solarwinds A2: In many ways. SSDs represent “throw hardware at the problem” which only works if you know disk is the issue. #datachat |
| jdanton | .@SQLRockstar @solarwinds I see a lot of storage bottlenecks. SSDs help with that lot. #datachat |
| SQLRockstar | RT @jdanton: @SQLRockstar @solarwinds But the SAN vendor sold me these magic beans with my SSDs. #datachat |
| SQLRockstar | .@jdanton @solarwinds My SAN vendor bought me sushi once. No magic beans though. Now I feel I got screwed. #datachat |
| jdanton | .@SQLRockstar @solarwinds It’s hard to tune a system where tempdb has 600ms write latency. #datachat |
| DBABullDog | @solarwinds @SQLRockstar A2 that Best application and database practices can be ignored and thrown out the window #datachat |
| SQLRockstar | .@jdanton @solarwinds Yep, when storage is the issue, SSDs can help. Agreed. #datachat |
| solarwinds | RT @josh_odgers: @jdanton @SQLRockstar SSDs in the wrong location can make no difference. Keep perf teir close to the app. #datachat |
| SQLRockstar | .@jdanton @solarwinds Chances are the disk itself isn’t the reason for 600ms latency. #datachat |
| jdanton | .@josh_odgers @SQLRockstar @solarwinds Good point–they are best treated as a limited valuable resource in any infra arch. #datachat |
| datachick | RT @solarwinds: Q2: What’s the number one mythunderstanding about SSDs? #datachat That their failure behaviours are same as spinny ones. |
| jdanton | @AirborneGeek @SQLRockstar @solarwinds I see locking, but yeah, storage is my most common. And it’s a mix. #datachat |
| SQLRockstar | RT @datachick: RT @solarwinds: Q2: What’s the number one mythunderstanding about SSDs? #datachat That their failure behaviours are same as … |
| solarwinds | RT @AirborneGeek As long as people haven’t done something dumb (like a 2-vCPU VM), disk is almost always the bottleneck, IME #datachat |
| solarwinds | RT @jdanton: @AirborneGeek @SQLRockstar @solarwinds I see locking, but yeah, storage is my most common. And it’s a mix. #datachat |
| josh_odgers | @solarwinds Adding SSD shelves behind a saturated central storage controller or storage area network wont help. #datachat |
| SQLRockstar | .@AirborneGeek @solarwinds I suppose it depends on the system, and if we are talking OLTP vs OLAP expectations. #datachat |
| jdanton | .@josh_odgers @solarwinds Great point–SAN configuration doesn’t get a lot of press, but it’s key for good perf. #datachat |
| SQLRockstar | .@AirborneGeek @solarwinds But IME a few code changes can make a huge improvement, quickly. #datachat |
| josh_odgers | @solarwinds If the NIC/HBA/Queue depths is the point of contention, an All flash array is not going to solve your problem. #datachat |
| jdanton | .@josh_odgers @solarwinds A fairly common approach in database servers is to use PCI SSD for temp space, which is highly random. #datachat |
| solarwinds | RT @josh_odgers: @solarwinds If the NIC/HBA/Queue depths is the point of contention, an All flash array is not going to solve your problem.… |
| solarwinds | RT @jdanton: .@josh_odgers @solarwinds A fairly common approach in database servers is to use PCI SSD for temp space, which is highly rando… |
| solarwinds | Q3: What are the key performance metrics you look at when troubleshooting VMs and why? #datachat |
| SQLRockstar | .@ToddKlindt @AirborneGeek @solarwinds The trick is to only read the page into the buffer once. That’s the real issue. #datachat |
| SQLRockstar | .@ToddKlindt @AirborneGeek @solarwinds SSDs are often just a band-aid to cover up design issues that won’t allow an app to scale. #datachat |
| jdanton | .@josh_odgers @solarwinds Getting data from all of those systems (SAN,Server and DB) in one view is a challenge. #datachat |
| josh_odgers | @jdanton @solarwinds local SSD shared logically across compute nodes in software (e.g.: #Nutanix) allows the best of both worlds. #datachat |
| ThugMetricsNews | RT @datachick: RT @solarwinds: Q2: What’s the number one mythunderstanding about SSDs? #datachat That their failure behaviours are same as … |
| AirborneGeek | .@SQLRockstar @ToddKlindt @solarwinds Well, sure, if you have more RAM than God. #datachat |
| jdanton | @SQLRockstar @ToddKlindt @AirborneGeek @solarwinds Or to provide IOPs without a ton of spindles that spinning media would require. #datachat |
| SQLRockstar | RT @AirborneGeek: .@SQLRockstar @ToddKlindt @solarwinds Well, sure, if you have more RAM than God. #datachat |
| jdanton | .@AirborneGeek @SQLRockstar @ToddKlindt @solarwinds I spec’ed a server last month with 792 GB of RAM. It was only $22k. #datachat |
| SQLRockstar | .@AirborneGeek @ToddKlindt @solarwinds I don’t need more RAM than God, just more RAM that the database size. #datachat |
| SQLRockstar | RT @jdanton: @SQLRockstar @ToddKlindt @AirborneGeek @solarwinds Or to provide IOPs without a ton of spindles that spinning media would requ… |
| DBABullDog | RT @jdanton: .@AirborneGeek @SQLRockstar @ToddKlindt @solarwinds I spec’ed a server last month with 792 GB of RAM. It was only $22k. #datac… |
| _Desdemona_ | RT @solarwinds: Q3: What are the key performance metrics you look at when troubleshooting VMs and why? #datachat |
| SQLRockstar | .@jdanton @ToddKlindt @AirborneGeek @solarwinds Fair point. But I like spindles. #datachat |
| SQLRockstar | RT @solarwinds: Q3: What are the key performance metrics you look at when troubleshooting VMs and why? #datachat |
| jdanton | .@solarwinds A3: In terms of SQL Server, I initially look at the same things I look at on physical. But I also look at 1/2 #datachat |
| SQLRockstar | .@solarwinds A3: VM CPU Ready Time is my favorite, followed by SQL Server wait events. #datachat |
| solarwinds | Q3: What are the key performance metrics you look at when troubleshooting VMs and why? #datachat . |
| jdanton | .@solarwinds A3: the number of CPUs (vCPU overallocation is bad). Also, want to look at other activites in vCenter/SCVMM. #datachat |
| SQLRockstar | .@ToddKlindt @AirborneGeek @solarwinds You know you can just use O365 instead, right? #datachat |
| SQLRockstar | RT @solarwinds: Q3: What are the key performance metrics you look at when troubleshooting VMs and why? #datachat . |
| SQLRockstar | .@jdanton @solarwinds A3: Oh yes, over-allocation is also something I look at right away. #datachat |
| DBABullDog | @solarwinds @SQLRockstar most cases I take the same approach as with physical. If answers are not provided start looking at host #datachat |
| jdanton | .@DBABullDog @SQLRockstar I had one case where PLE suddenly dropped dramatically. The memory wasn’t reserved and VMW took it. #datachat |
| SQLRockstar | .@jdanton @DBABullDog Lemme guess, no LPIM set? Min memory not set? #datachat |
| datachick | RT @solarwinds: Q3: What are the key performance metrics you look at when troubleshooting VMs and why? #datachat . < Cries from users. |
| SQLRockstar | RT @datachick: RT @solarwinds: Q3: What are the key performance metrics you look at when troubleshooting VMs and why? #datachat . < Cries f… |
| jdanton | .@SQLRockstar @DBABullDog LPIM is a different story. I don’t always recommend it. cc @mrdenny #datachat |
| SQLRockstar | .@datachick @solarwinds Cries from users? I once thought about sending shocks to user keyboards. #datachat |
| SQLRockstar | RT @jdanton: .@SQLRockstar @DBABullDog LPIM is a different story. I don’t always recommend it. cc @mrdenny #datachat |
| datachick | RT @AirborneGeek: .@SQLRockstar @ToddKlindt @solarwinds Well, sure, if you have more RAM than God. #datachat |
| jdanton | .@SQLRockstar @DBABullDog Reserving at the hypervisor, is what I usually recommend. #datachat LPIM can have some bad OS impacts. |
| SQLRockstar | .@jdanton @DBABullDog @mrdenny I used to recommend LPIM on VMW until recently I was persuaded to change my mind slightly. #datachat |
| jdanton | .@SQLRockstar @DBABullDog @mrdenny Also, never set min/max memory to the same value. #datachat |
| DBABullDog | RT @SQLRockstar: .@jdanton @DBABullDog @mrdenny I used to recommend LPIM on VMW until recently I was persuaded to change my mind slightly. … |
| SQLRockstar | .@jdanton @DBABullDog @mrdenny Ah, there are two cases where you *might* want min=max. One case is multi-instance. #datachat |
| SQLRockstar | .@jdanton @DBABullDog @mrdenny I forget the other case…more scotch might jog my memory. #datachat |
| jdanton | @SQLRockstar @DBABullDog @mrdenny There is this though. https://t.co/Gv3upb8Qyj #datachat |
| solarwinds | Q4: Do SSDs and flash arrays solve everything? #datachat |
| jdanton | .@SQLRockstar @DBABullDog @mrdenny Multi-instance is usually better addressed with startup triggers, or something. #datachat |
| jdanton | .@solarwinds A4: If everything is taking a lot of money out of you account, yes. They are really good at a lot of things. 1/2 #datachat |
| SQLRockstar | .@jdanton @DBABullDog @mrdenny Yeah, that’s one of the BOL entries I asked to be updated to not say min=max. #datachat |
| josh_odgers | “@solarwinds: Q4: Do SSDs and flash arrays solve everything? #datachat” < Not without solid supporting infrastructure and app architecture |
| jdanton | .@solarwinds A4:They stil have SAN problems like mentioned earlier (controller overload etc),but they offer excellent performance. #datachat |
| SQLRockstar | .@jdanton @DBABullDog @mrdenny There’s also the fact that SQL Server “paging” is not the same as Windows O/S paging. #datachat |
| SQLHA | @solarwinds SSD/Flash is not magic. It can help, but not in all cases and will not solve issues of bad app. ONly mask longer. #datachat |
| SQLRockstar | .@jdanton @DBABullDog @mrdenny Saw a great talk by @bobwardms at #sqlbits last week, he talked about LPIM, min = max, paging, etc. #datachat |
| SQLRockstar | RT @solarwinds: Q4: Do SSDs and flash arrays solve everything? #datachat |
| solarwinds | RT @SQLHA: @solarwinds SSD/Flash is not magic. It can help, but not in all cases and will not solve issues of bad app. ONly mask longer. #d… |
| jdanton | .@SQLRockstar @DBABullDog @mrdenny Right, but it can lead to hypervisor paging, which is REALLY bad for performance. #datachat |
| solarwinds | RT @jdanton: .@solarwinds A4:They stil have SAN problems like mentioned earlier (controller overload etc),but they offer excellent performa… |
| SQLRockstar | .@solarwinds A4: No, they don’t solve bad code and design, nor do they solve the PEBKAC issue. #datachat |
| SQLRockstar | RT @SQLHA: @solarwinds SSD/Flash is not magic. It can help, but not in all cases and will not solve issues of bad app. ONly mask longer. #d… |
| SQLHA | @jdanton @SQLRockstar @DBABullDog @mrdenny Which is why you monitor at hypervisor and in guest to get whole picture. #datachat |
| jdanton | .@solarwinds A4: Another intestesting thing is that many all flash arrays have compression. It tends to work with DB compression. #datachat |
| SQLRockstar | .@SQLHA @jdanton @DBABullDog @mrdenny Exactly! You need the WHOLE picture, not just one piece. #datachat |
| SQLRockstar | RT @jdanton: .@solarwinds A4: Another intestesting thing is that many all flash arrays have compression. It tends to work with DB compressi… |
| SQLRockstar | .@jdanton @solarwinds Oh? You can double-compress? #datachat |
| jdanton | .@SQLRockstar @solarwinds The algos on the array sides are pretty light. I talked to a colleague at a major vendor and he 1/2 #datachat |
| DBABullDog | @solarwinds cases I have observed something like this is short term gains. Eventually high performance apps outgrow HW gains. #datachat |
| jdanton | .@SQLRockstar @solarwinds did some extensive testing, and at worst case, there were no compression. But it never got bigger. #datachat |
| SQLRockstar | RT @DBABullDog: @solarwinds cases I have observed something like this is short term gains. Eventually high performance apps outgrow HW gai… |
| DBABullDog | RT @SQLRockstar: .@SQLHA @jdanton @DBABullDog @mrdenny Exactly! You need the WHOLE picture, not just one piece. #datachat |
| solarwinds | Q5: What are some of the challenges you have when working with SSDs? #datachat |
| SQLRockstar | RT @solarwinds: Q5: What are some of the challenges you have when working with SSDs? #datachat |
| josh_odgers | “@solarwinds: Q4: Do SSDs and flash arrays solve everything? #datachat” < Adding SSD/AFA can simply highlight the next bottleneck ie CPU/Ram |
| jdanton | .@solarwinds Not SSDs per se, but tiered SANs (which include SSDs) offer some unique challenges. #datachat |
| jdanton | .@josh_odgers @solarwinds Or drive more CPU use just from throughput. #datachat |
| SQLRockstar | .@solarwinds A5: Getting people to understand that HA <> DR. Without a proper DR plan, if the SSD array fails, you are in trouble. #datachat |
| josh_odgers | “@jdanton: .@josh_odgers @solarwinds Or drive more CPU use just from throughput. #datachat” < Yes! remove disk bottleneck & cpu can increase |
| SQLRockstar | .@solarwinds A5: Why buy one SSD when you can buy two at double the price? #datachat |
| jdanton | .@solarwinds A5: Other challenge is convincing clients that it’s not the answer to all problems. #datachat |
| DBABullDog | @solarwinds @SQLRockstar A5 having all parties involved learn about the technology being implemented. In this case SSDs. #datachat |
| SQLRockstar | RT @josh_odgers: “@solarwinds: Q4: Do SSDs and flash arrays solve everything? #datachat” < Adding SSD/AFA can simply highlight the next bot… |
| jdanton | .@SQLRockstar @solarwinds Wait, SSDs need RAID? #datachat |
| DBABullDog | RT @SQLRockstar: .@solarwinds A5: Getting people to understand that HA <> DR. Without a proper DR plan, if the SSD array fails, you are in … |
| AirborneGeek | RT @jdanton: .@SQLRockstar @solarwinds Wait, SSDs need RAID? #datachat |
| SQLRockstar | .@jdanton @solarwinds Not if the array has it already, likely software defined. But still, you need a DR plan. #datachat |
| SQLRockstar | .@jdanton @solarwinds And RAID is not enough to satisfy a proper DR plan. Not if I’m involved. #datachat |
| SQLRockstar | RT @jdanton: .@solarwinds A5: Other challenge is convincing clients that it’s not the answer to all problems. #datachat |
| SQLRockstar | RT @DBABullDog: @solarwinds @SQLRockstar A5 having all parties involved learn about the technology being implemented. In this case SSDs. #d… |
| jdanton | .@SQLRockstar @solarwinds Yeah, I agree completely–a good example would be an Availability Group where you had SSD on Primary 1/2 #datachat |
| josh_odgers | “@solarwinds: Q2: What’s the number one mythunderstanding about SSDs? #datachat” < Adding performance without data services can reduce value |
| rajesh_dangi | RT @solarwinds: Q5: What are some of the challenges you have when working with SSDs? #datachat |
| jdanton | .@SQLRockstar @solarwinds and spinning on secondary–index rebuilds won’t go so well there. #datachat |
| josh_odgers | “@solarwinds: Q5: What are some of the challenges you have when working with SSDs? #datachat” < Client perception it will solve all problems |
| solarwinds | RT @jdanton: .@SQLRockstar @solarwinds and spinning on secondary–index rebuilds won’t go so well there. #datachat |
| solarwinds | RT @jdanton: .@SQLRockstar @solarwinds Yeah, I agree completely–a good example would be an Availability Group where you had SSD on Primary… |
| jdanton | .@sqlha Brings up a good point–SSDs aren’t really good at sequential writes. Unless firmware fixes that. #datachat |
| DBABullDog | RT @jdanton: .@sqlha Brings up a good point–SSDs aren’t really good at sequential writes. Unless firmware fixes that. #datachat |
| jdanton | Most of the major SAN vendors have great firmware to make sequential write random, but something to think about. #datachat |
| vcdxnz001 | RT @SQLRockstar: .@solarwinds A5: Why buy one SSD when you can buy two at double the price? #datachat |
| SQLHA | @jdanton And which slot, say, you put an FIO card into matters, too. Not all slots are the same speed. #datachat |
| SQLRockstar | .@jdanton @sqlha I don’t think SSDs would ever want sequential writes, nor should they be needed? #datachat |
| solarwinds | Q6: If the hypervisor were a superhero, which one would it be and why? #datachat |
| jdanton | .@SQLRockstar @sqlha Well in an all flash array. And the IOPs still scale. The writes effectively become random with firmware. #datachat |
| SQLRockstar | RT @solarwinds: Q6: If the hypervisor were a superhero, which one would it be and why? #datachat |
| jdanton | .@solarwinds A6: Stretch Armstrong–the hypervisor has to cover a lot of ground. #datachat |
| AirborneGeek | .@solarwinds Which one holds up the entire world? #datachat |
| datachick | RT @solarwinds: Q6: If the hypervisor were a superhero, which one would it be and why? #datachat < Darth Data. She’s the best. |
| SQLRockstar | RT @datachick: RT @solarwinds: Q6: If the hypervisor were a superhero, which one would it be and why? #datachat < Darth Data. She’s the be… |
| SQLRockstar | RT @jdanton: .@solarwinds A6: Stretch Armstrong–the hypervisor has to cover a lot of ground. #datachat |
| YvesMulkers | RT @SQLRockstar: .@solarwinds A5: Why buy one SSD when you can buy two at double the price? #datachat |
| SQLRockstar | RT @jdanton: .@SQLRockstar @sqlha Well in an all flash array. And the IOPs still scale. The writes effectively become random with firmware.… |
| solarwinds | RT @datachick: RT @solarwinds: Q6: If the hypervisor were a superhero, which one would it be and why? #datachat < Darth Data. She’s the be… |
| jdanton | .@datachick @solarwinds Yes, but what was her super power? #datachat |
| josh_odgers | @solarwinds storage controllers also need to be scaled to support more SSD. #Scaleout #datachat |
| datachick | RT @solarwinds: Q6: If the hypervisor were a superhero, which one would it be and why? #datachat < Prof. Charles Xavier. |
| jdanton | .@josh_odgers @solarwinds Most vendors I’ve worked with do a good job with that. Are you seeing issues with a particular config? #datachat |
| SQLRockstar | .@solarwinds A6: Charlie Townsend, he controls it all but you never see him and only speak with him through a dedicated interface. #datachat |
| SQLRockstar | RT @jdanton: .@datachick @solarwinds Yes, but what was her super power? #datachat |
| SQLRockstar | RT @datachick: RT @solarwinds: Q6: If the hypervisor were a superhero, which one would it be and why? #datachat < Prof. Charles Xavier. |
| DBABullDog | RT @datachick: RT @solarwinds: Q6: If the hypervisor were a superhero, which one would it be and why? #datachat < Darth Data. She’s the be… |
| SQLHA | @solarwinds Batman. Has to be the smartest guy in the room and handle every situation. #datachat |
| SQLRockstar | RT @SQLHA: @solarwinds Batman. Has to be the smartest guy in the room and handle every situation. #datachat |
| SQLRockstar | .@SQLHA @solarwinds Same could be said for Tony Stark. #datachat |
| SQLHA | @SQLRockstar @solarwinds Marvel vs. DC. DC wins. #datachat |
| SQLRockstar | RT @SQLHA: @SQLRockstar @solarwinds Marvel vs. DC. DC wins. #datachat |
| SQLRockstar | .@SQLHA @solarwinds DC wins? Because Superman? #datachat |
| SQLHA | @SQLRockstar @solarwinds And the hypervisor is supposed to be more invisible. Tony Stark is a showoff .Batman is stealthy. #datachat |
| jdanton | An aside from tonights #datachat, but I’d like to raise a glass to @gigaom, we’ll miss you. |
| SQLRockstar | .@datachick @solarwinds Wha? Dude in a wheelchair running a school for rejects? That’s not like a hypervisor, not at all. #datachat |
| solarwinds | Q7: And, if you were the Academy, would the Oscar go to Hyper-V or VMWare? #datachat |
| DBABullDog | @SQLHA @SQLRockstar @solarwinds I always get the feeling the hypervisor is showing off. #datachat |
| SQLRockstar | RT @SQLHA: @SQLRockstar @solarwinds And the hypervisor is supposed to be more invisible. Tony Stark is a showoff .Batman is stealthy. #data… |
| AirborneGeek | .@solarwinds A6. wait. Going about this all wrong. Jean Luc Picard is a superhero, right?Because, obviously #datachat |
| DBABullDog | RT @jdanton: An aside from tonights #datachat, but I’d like to raise a glass to @gigaom, we’ll miss you. |
| SQLRockstar | RT @DBABullDog: @SQLHA @SQLRockstar @solarwinds I always get the feeling the hypervisor is showing off. #datachat |
| SQLRockstar | RT @AirborneGeek: .@solarwinds A6. wait. Going about this all wrong. Jean Luc Picard is a superhero, right?Because, obviously #datachat |
| SQLRockstar | RT @solarwinds: Q7: And, if you were the Academy, would the Oscar go to Hyper-V or VMWare? #datachat |
| SQLHA | @solarwinds Neither if we’re being honest. They’re more similar than different. #datachat |
| DataDefiner | Darth Data is my father. |
Leave a Reply