Pass Summit 2016 abstract reviews

Following on Steve’s blog post on summit abstracts, I decided to publish mine.

My comments are not intended as an attack on the program committee, it’s an incredibly hard job that you couldn’t pay me to do (I’ve done similar work before and hated it). What I hope comes out of this, and the other posts which Steve reviewed, are that comments are more constructive and can be used to improve the abstract.

When the initial mails were sent out, I had one general session and one lightning talk accepted. Since then, I’ve been asked to also present one of the sessions that was listed as alternate.

So, without any editing, the abstracts and comments:

All about Indexes (half-day, level 300, declined)

Indexes are essential to good database performance, but it can be hard to decide what indexes to create and the 'rules' around indexes often appear to be vague or downright contradictory.

In this session we'll dive deep into indexes, have a look at their architecture and internal structure and how that affects the way that indexes are used in query execution. We’ll look at why clustered indexes are recommended on almost all tables and how their architecture affects the choice of columns. We’ll look at nonclustered indexes; their architecture and how query design affects what indexes should be created to support various queries.

Comments

Abstract starts off really well, but last sentence has some structure/repetition issues. Level might be appropriate, but I wonder if in a 3 hour session you can cover enough detail to justify a 300 level.

Nice abstract and overall flow between title, abstract and goals.

Abstract seems well written and gives decent insight into session contents. It would be better to mention 2014 & 2016 features to make the session more attractive. The topic of indexes is somewhat overdone and may have some difficulty attracting attendees. Goals are decent but somewhat generic and could benefit from being more tangible. 25% demo on this topic seems a bit low to keep attendees engaged for a 3 hour session.

Need more meat in abstract. Are you going to talk about column, filtered or other index types? Only 25% demos, you have 3 hours???

Could cover more breadth in 3 hours. Filtered, sparse, columnstore etc.

I don’t see this for a half-day session. The goal are just too weak. And only 25% demo’s? You mean you are going to be talking for 3 hours and demoing for an hour? zzzzz. Need to flesh this abstract out a lot more.

Subjective: For a 3h long session, even at lvl300 , would expect a bit more demos to keep the attendees from nodding off…

Interesting combination. One wonders if I can cover enough material in 3 hours to make it a level 300 and others imply that there’s not enough material to spend 3 hours on.

Thing is, I’ve given this presentation twice before, as a 3 hour session. Pass Summit in Charlotte and SQLBits in London. It fits into 3 hours, providing there aren’t too many questions, that is. If I was going to add clustered and nonclustered columnstores and Hekaton’s range and hash indexes (and the interactions between them, eg non-clustered rowstore index on a clustered columnstore), then it would be a full day precon at least.

There are no 2014 or 2016 features mentioned, because the rowstore indexes didn’t change much in either version. Sure, I could call out the increased key size, but that’s not exactly abstract material. Maybe a footnote on one slide.

As for the topic being overdone and will have difficulty attracting attendees, there are only two sessions on rowstore indexes this year, a precon (Kendra Little) and a 100-level general session (Kathi Kellenberger), plus a few internals presentations and performance-related presentations that include indexing. Every time I’ve given an indexing presentation at a SQLSaturday, Pass Summit or SQLBits I’ve had a packed room. Sure, it isn’t shiny and new, but it’s probably relevant to most systems we design, develop and administer.

“You mean you are going to be talking for 3 hours and demoing for an hour? “ <snark_mode = on> No, I planned to do the middle hour as a mime act. </snark_mode>

Last time I checked, a 3 hour session meant talking for 3 hours. I’m sorry that indexing sounds so dry and boring that attendees are sure to fall asleep during it. I haven’t yet had anyone falling asleep in my indexing sessions. Maybe everyone who did, did so quietly and I didn’t notice. Maybe I should bring fireworks next time.

Go, Go, Query Store! (general session, level 200, accepted)

One of the hardest things to do in SQL is to identify the cause of a sudden degradation in performance. The DMVs don’t persist information over a restart of the instance and, unless there was already some query benchmarking (and there almost never is), answering the question of how the queries behaved last week needs a time machine. Up until now, that is. The addition of the Query Store to SQL Server 2016 makes identifying and resolving performance regressions a breeze.

In this session we’ll take a look at what the Query Store is and how it works, before diving into a scenario where overall performance suddenly degraded, and we’ll see why Query Store is the best new feature in SQL Server 2016, bar none.

Comments

Topic: Great topic. SQL Server 2016 and new shiny features like QS will definitely draw a crowd.
Abstract: Very well written with clear supportive goals.
Subjective: This sounds like a great session I would like to attend. The one issue I have is that it is listed as only 25% demo. Demos are the best way to who off QS functionality. 50% demo would be better.

I would like to attend this session

The outline seems to clearly describe the contents of the presentation. The topic and goals should be compelling to attendees. The target audience should be big enough to support this session. There appears to be a reasonable amount of live demonstrations in relation to the topic being presented.

Abstract: detailed, clear
topic: relevant and new, one of the more interesting feature of sql server 2016
Subjective rating: a good session

Abstract: well written, good topic
Topic: good topic, eye catching
Subjective: good session, would be interested in this

Nothing much to say here. The session will likely be closer to 50% demos, the original plan was 25%, but when I built the slides for SQL Saturday Iceland, I ended up putting more demos in and less slides.

Is that a parameter I smell? (general session, level 300, declined)

All too often a forum post on erratic query performance is met with a reply &quot;Oh, it&#39;s parameter sniffing. You can fix it with &lt;insert random solution here&gt;.&quot; The problem with answer, even if it has identified the cause, is that it’s only part true. Parameter sniffing is not simply a problem that needs fixing; it&#39;s an essential part of well-performing queries. In most cases.

Come to this session to learn what Parameter Sniffing really is and why it’s a good thing, most of the time. Learn how to identify the scenarios where it’s not good, why a feature that is supposed to improve query performance sometimes degrades it, and what your options are for resolving the problems when they do occur.

Comments

Abstract: Well presented, explains purpose of session.
Topic: catchy title, still timely and relevant.
Subjective: Nice mix of demo for this topic.

title not cute, if someone does not know about this they would not come Good abstract and good goals

Abstract: Grammar is a bit rough making the abstract difficult to read.

topic is decent . while it not “latest” or “hot”, it address a real life problem that developers can face. Abstract is a bit too generic and could benefit from more tangible details. Goals are okay but somewhat generic. demo % is decent.

Thanks for submitting a session that covers a really hot button topic in database queries! Looking forward to seeing it.

Again, fair enough, nothing much I can say about these comments.

On Transactions and Atomic Operations (general session, level 200, declined)

If there’s one thing that we, as SQL developers, do, it&#39;s not use enough transactions.

Transactions are critical when multiple changes need to be made entirely or not at all, but even given that it’s rare to see transactions used at all in most production code

In this session, we&#39;ll look at what transactions are and why we should use them. We&#39;ll explore the effects transactions have on locking and the transaction log. We&#39;ll investigate methods of handling errors and undoing data modifications, and we&#39;ll see why nested transactions are a lie.

Comments

Abstract: Grammar in first sentence is difficult to read. Too many commas.

good topic, need more meat in abstract and goals.

topic may not be “latest” but its a foundation subject and of relevance to most database developers. abstract clearly conveys what to expect from the session. goals are somewhat terse and could benefit from tangible details. demo % is a bit low.

Atomic Operations is in the title, but not used/defined in the abstract itself.

Abstract: Second sentence is a bit run-on and incoherent.
Topic: Fit for purpose, but likely a bit niche, as the author already seems to fears.
Subjective: Would have expected at least 50% demo – after all the easiest way to understand the need for TRANS is to see data buggered up by colliding updates.

Yes, I do have a tendency to torture my commas (they always confess at the end) and play with grammar. I will try to remember next year to write with simple, plain grammar. I do need to flesh this out a bit for next time it’s submitted anywhere.

I’m not sure what in the abstract conveys my fear that it’s a niche session, I don’t think it is. From working on various clients’ production systems over the last few years, I very seldom see explicit transactions. It’s something we don’t do enough, not a niche topic.

I’m also not sure that the reviewer meant by ‘colliding updates’. Updates, like all statements, are always part of a transaction, even if it is just one automatically started and committed. If two sessions were to try and update the same column, same rows at the same time to different values, the outcome will be as if one or the other had run, not a mixture of the two. And wrapping the update in an explicit transaction won’t change that behaviour

The Many Latencies of TempDB (general session, level 400, originally alternate, later accepted)

TempDB gets a bad rap when it comes to performance and scalability, and it’s all-too-often well deserved. A badly configured TempDB can have devastating effects on throughput. Combine that with poor queries and, well, you didn’t have any plans for the weekend, right?

In this session we’ll look at some common causes of TempDB contention, both query-based and configuration-based. We’ll look at guidelines for configuring TempDB and when and why you’d make various changes, and we’ll cover more ways to monitor TempDB than you can shake a stick at.

Now, about those weekend plans…

Comments

Abstract: Clear and well written abstract. No prerequisite listed.
Topic: Topic and goals seem like they would be appealing to attendees.
Subjective: Sounds like a good session that people would benefit from.

Abstract: interesting
Topic: relevant
Subjecttive rating: compelling, high level

Abstract: The outline and details of this abstract are well written
Topic: This is a good topic
Subjective: I may attend this session

The session prerequisites are helpful. A great topic and the Abstract is almost good. It is funny but not enough information about the session.

Session prerequisites: TBC?
Goal 1: Correct the “&#39;”
Objective: I would like to attend this session.

The pre-reqs here were a mistake on my part. Despite checking and re-checking the abstracts, that mistake slipped through (tbc = to be completed).

I would love to have corrected the “&#39;”, as well as the &quot; and &gt; that appeared scattered through the abstracts, but they appeared on submit and editing afterwards didn’t help. Something, somewhere in the website messed up the HTML encoding a bit.

Watch a query run. Run, query, run! (Lightning talk, level 100, declined)

Previously if you wanted to get any run-time statistics for a query, you had to include the actual execution plan and run the query to completion. No more! New in SQL 2016 is the live query statistics that let you watch the execution of a query, in real time, and see when the operators run and where the data flows.

In this lightning talk we’ll look at how the live query statistics works and discuss some scenarios where this will really help debugging strange query behaviour.

Comments

Perfect topic/abstract for Lightning Talk, sign me up!

interesting and current topic. well written abstract with good details of session contents. Goals are terse but clear.

Sounds like a good topic for a Lightning Talk.

Great topic and good intro to query store. Hopefully this talk will provide concise knowledge.
Topic is good, will interest a large amount of attendess.
Level is excellent – it’s a new feature so 100 level is great. I would have liked to see mention of query store in the title.

good topic

New topic on new technology. Small enough to be interesting

Interesting and great topic. Abstract and Goals talks to each other. New features is always a good place to grab peoples attention.

The comment on wanting query store to appear in the title is mystifying, because this session has nothing to do with query store. It’s showing off the Live Query Statistics (which, while nice, is too small to warrant anything more than a lightning session alone)

Why Temporal Tables and Stretch Database Are Best Friends (Lightning talk, level 200, accepted)

SQL Server 2016 introduced, among several other new features, Temporal tables and Stretch database.

From a distance, they appear not to have much to do with each other. Temporal tables allow you to query the table as it was at a point in time. Stretch allows for tables to span the earthed SQL Server and the cloud.

In this short session we’ll look at why these two features work spectacularly well together and why a stretched temporal table makes perfect sense.

Comments

Clear understanding of the objective, and as a “new” feature people will drawn to this.

Perfect Lightning Talk abstract, focused topic and something new with the latest version of SQL Server.

Sounds interesting!

interesting topic – new and relevant. Abstract is brief but gives good insight into session contents. Goals are terse, but clear and tangible. No demos may be off-putting to some attendees.

Earthed SQL is an interesting expression and certainly made me think.
There’s two new features here – is that too much for a 10 minute lightening talk?
If good connection is made early between the two then this could be good
Three goals seem to sum up the presentation well.

Try just one topic in 10 minutes or create a 75 or 3 hour session on new features of 2016

No real “pull” for a short session.

“No demos may be off-putting”. Um, what? This is a 10-minute lightning talk, I’m not demoing features in 10 minutes and I’d be very surprised if attendees expected demos in a 10 minute session.

I can’t take credit for the “Earthed SQL” expression, that one’s from Rimma’s Keynote presentation in 2014