Merken

Visualizing Postgres in realtime

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
both of them are and so but rise of visualizing of claims will whatever
work on Africa for
other proposed 1st talk about
their request resting but before really get into this on the talk little bit about the talk itself and a lot of the work and look at the schedule for the other talks I saw that a couple of things that I was tending to cover was very pretty well covered by such as generating stuff along streams and doing advanced in data storage role of that and so on but before I do that I had you know put our now and talk things I want to talk about it and I looked at and those kind of boring so I take up the parts that here I thought were kind of boring and when he thought about all the talks had seen in the last couple years and which ones are remembered and which ones are I not follow the most interesting and a lot of them weren't super technical the warrants are full of code they were more of your cover general topics that I was interested here because you know in a programmer nerve but wasn't really specifically you code or maybe not engineering and so you know that to the topic of a woman talking individuals and processes about is a perfect opportunity to go into sort of the history of datavisualization 10 sigh I when research that was actually a really interesting so I hope you'll find it as interesting as I have found don't work as talkers on all get more and more specific and more practical but I'm a start out with a short history lesson I but if you
forget to that but I want to look at you know why do we want to visualize data in general you know it's just take granted the colony demographic spread this really good reasons why you would want to do any good data out
1 of this this first one is sort of the Apollo attract but it does illustrate the point of this guy had made these 4 sets of numbers slack pairs and you look at the numbers and just like OK that's that's sort of so let's run some statistical analysis on and select the 2nd tallest more about the sets of numbers it turns
out that each 1 of those pairs has the same for the exit has the same mean for the why it has the same meaning both at the same variance the correlation between suppose force that's exactly the same and the linear aggression so I ask all those 4 sets are the same
Boeing graphic you'd see that they're completely different I have no idea if you had you a more broad set of disco things you probably see that there but it is looking at that really brings home the point of why you know it's useful to legislative notices here's the sort of statistical poetry in but here for data stuff as engineers operations and the
main thing I think that this is can help for is in times of crisis are page and overnight persistent going down but I pili whatever sent that page it doesn't have enough of that information it's in the page should have enough information to solve otherwise you could just automated and many of them around on many but you know for the interesting problems that require no human intelligence or whatever so the page is enough but if you have a good graph of the system nice charts and you can see at a glance how much faster what's wrong than dropping through lots finding out all the different parts were large system that's a sample or other system logs like trying find and in that in that sense but having good position the system can reduce can lead to reduced down which is a good thing are related to
that as so well as 1 thing
that this is a system we have the stairwell so there's an outage with light lights up the stairwell because it is like that like go up the sides of walls this was actually a of running like this and this report from another thing related
to crisis is is just anomalies in general but maybe it hasn't maybe something's wrong wrong you haven't seen before that you know you have a page you can notice some odd behavior the can prompt investigatory work to see if so here is so that's about to go wrong and the reason you can do that is because you can start seeing with visual systems the interactions between lots of complex systems more than because you don't have to know what you're looking for ahead of time you can see the strange or
another good reason visualizes for capacity so you can tell if you should upgrade soon you know you can look at cycles over days or weeks and see if you perhaps in the middle of the day you always have a busy period and so on and so forth and that can increase about the robustness of but really altogether it
removes like gut feeling guessing puts your working systems into a more and more signs a III has if you scientific but more data-driven you know approach
so my have 2nd talk is the history I think I know you can if you looked at things that people have done in the past you can get you get inspiration you can give you lessons of what you know what they did and I think like some of the stuff when I actually turned into a structured prediction so the 1st guy
is William Playfair so indeed he was in the late in the early 18 hundreds and
he made the the 1st like line chart the 1st paragraph 1st bar but before him like but they had no graphs but no 1 used them to show data and try and you know you explain stuff visual manner and so this is 1 of the of as the 1st line chart but it's 1 of the 1st line this shows a tree balances between Norway and Denmark and England and this doesn't look that strange because the stuff that he did everyone ended up using you know forever like this title at the top there's a few frame around the graph the axes are labeled it has a caption has good use of color and I I 1 of the interesting things is site in the side of the book was published explanation of how to read the chart itself like I you know if you want to see the exports from 1940 father 1740 follow the line from 1740 up and over to the right to find out what it's like he like had to write it out in words how these because it wasn't you know commonplace that I just
added a bunch of other crazy line charts there like that you cursive qualification type title of the song was showing during wars with the French all of these other metrics use tracking were kind of normal but then after these 2 wars of the top inserted in his opinion going like we have control has been this is the 1st
part shot that is made this is interesting because the the previous 2 of all time series that the
head of you know the years and going on about this was
his 1st graph that compared things without the aspect of time on the bus and again you know this is of bar chart is still recognized today and and it's you know he did the
1st pie chart that this was showing an at population sizes and how much their their government tax them and this is mostly good there's 1 thing that's misleading about this charge so which is unfortunate is you draws these bars on the top and then showing to the numbers and in connection with a lot but the the slope of that line you can see all the slopes the slope doesn't mean anything from the data just because of how wide the circle and you get out of the absolute sign of that shows that there's more taxes application of the slope doesn't mean that I unfortunately I guess people you even from the beginning that something's wrong with it and think through and when exterior guy is
Johnson it's not that think I see is a position in London and there was
this sack cholera epidemic this huge cholera outbreak and so at the time the germ theory was known but not widely accepted and that's the theory that there exist germs that get you sick and if you accept germ 3 then you might wash your hands are clean things before you do the surgery and so there was this that epidemic of cholera and he you know service suspected that you media sources and so he had started tracking where people died you know black lines
and he notices that you know I have a lot of room around this
the center and bots Broad Street and then turns out there's a a water pump that got contaminated and everyone who's drinking from that got sick and died out what would happened was is they had cesspits at the time which were actually just underneath your house where you know sort of like a not very good at a septic tank and that 1 of them wanted some future about cholera and diapers when sees the thing that leaked into the well they didn't and so this the with this analysis they figure out that that was the pump and the government claimed that they'd never at the time he gave him credit because the thought of some landscapers in the water was too outlandish for the time and so it until later that everything was itself has to unitary now there's actually can pump statues commemorating and so he did you needed sequential lights on there was 1 also from part of the story is that there was a monastery nearby as a brewery and none of the monks that monastery got sick and that's because they didn't drink the
water filling drink the year that it themselves so that's graph has the
nice guys that Charles Joseph is has also contribution to the relations and that is
this chart which has 0 that said this is the office but it needs if you haven't seen before it needs explanation but the density of how much is there is just as great Edward Tufte said it might be like the best graph of sort it is is the invasion of Napoleon's invasion of Moscow and then retreat and with so you can just see at the left here so that the the width of the line of how many troops are at that point they go forward to forward to Moscow and then and then back and you can just
see like how devastating invasion was just by comparing the start out with and what came back and they started with it says here foreign 22 thousand came back so at every point you can see it so I you know we have a branch break off to protect in some other area they go forward in a unit of the
part of the dilution shoots it Moscow region around the black line back in the black line back at the bottom here the track the temperature and this winter was freezing freezing cold other units are here and I sat sources since the mind will so solution are not use that temperature yeah and which pioneered the conversion celsius glycolysis office and so like this this line here starts at 0 and then just go it's called a negative 30 and so the lines here on map the temperature and every time you know the temperatures down the more people get lost in this sort crossings river you can tell it's just off like so it's a fossil that this graph has like you know so much was like look at it like this you the story of you know what have I but then this
would be complete without that repetitive are the mapping a modern-day contemporary
realization expert for books and I recommend you go out there and get these if any of this is interested in the which really great about these boxes that wanted to use a big believer in having a very dense high bandwidth information and he couldn't find a publisher that would print books the way that he wanted so he made his own publishing company in the paper on flip through these books is just like a really thick and colors are all in a very great I and so I I was better than reading through the preparation of stock re read through all of them in reading through a couple the other ones and I wanna show a couple of good ideas that I thought were about some of the most valuable to doing something for visualization of the stress
and 1 thing that he talks about a lot is that data in creation of a graph to need some particular growing the use printer graph that some of it goes to displaying information and some of it goes to like the frames or the good ones rover and his idea is that you should minimize to as much as it can the actually that doesn't actually give any information so like if you have a really fat Barbara but you know that the with actions and signifying anything but that's not too much because you only need like how high how high and things like that but another
thing he goes on a lot about is making sure that you're visualizations are honest and some people know intentionally mislead with you know trying to distort what's going on but even more subtle than that I know I've misled with graphs by not including time zone and that just you know like you don't know if this was is this specific pharmacists UTC it's probably PC but we share and I that just causes so much you know intentionally being dishonest not intentionally being assigned this sounds bad but still and 1 last thing is
is that this is a random sentences in a big point about that is the great that information consistent differences that make a difference you can have a lot of noise in year in year visualizations is if you minimize that just keep the 2 most important differences then you can get your point across much faster and much much more clarity and the thing that remains
are brought him to my attention 1st places that made spark on the of the 1st time I saw a such a way that maximizes these useless but I don't understand the point until I actually read through his analysis of SPARQL lines of of the site and you know is actually posted that part of this chapter for free you can read through and that this is an example from that I can get it pretty but what it's showing is that index funds versus mutual funds and you can see that over time they all units of that was just which awful from approach the same return and so the server proves a point that index from vantage points because they have a lot of this ratio they all perform efficient and the what the women do that is by having his idea small-multiples and so it's the same graph several times and so you can compare the differences between them much better than if each 1 of these were like in a grid of 6 you can really line them up because they weren't is it a small multiples of so I
went been doing some relations on that was present in 1 type of chart that I think everyone causal when there's another 1 that I found that I'd like a lot that's a fairly common and so that you know whatever knows the line
charge and this is a graph from a service called the broader uh injustices antifibrotic but I'm so brothers pretty awesome itself hosted metrics collections of sequence and metrics to I I sort of love-hate relationship but it does its job very well but it's just it's sometimes hard to work with them I think they get a couple things wrong but we use it for a lot of stuff internally because it's so we don't have to build our own stuff for working this is showing is such that he so another reason about as negatives so it's hard to read on projectors and its British have a laptop but I don't need it on so merits those lines are but 1 of the things that I don't like about it is because it is a grid of 6 things you can't line up at the current time and so you have to go through and you can see if something is related to this is if you were abroad coming can see this example that's alive like data from the regions bombed project that we run on since the loss softly rather and they have but you know the the taking some stuff from the stress here that posters and they their follower I read replicas the database what the cash reason that what I'm having index cancer happening and the sequences of the sequential scan so and that that's OK so I but
1 of the things that you have to make sure that you do before you start showing data line graph is you're probably reducing data down to 1 or 2 numbers to tests that online and the natural thing to do is to just do the average but a lot of times that's completely wrong but 1 of the much better things to do if you look at before you make your graph looked at the history of the data and see if it happens to be normal like OK the news is just an average but a lot of times I've seen when you actually get the history of the data but be a lot of like really quick things like what they doing a web requests this is a really important 1 for not using the average response time because a lot of your requests are going to be very fast but then a handful requested to be very long and that of by moral sort of thing that makes the average not representative of what you really want to look at % 95 a % 99 response time to get a better idea of how the system's action on so this this next graph is actually my favorite and you can see uh there's there's a paper introducing the goes into detail of how to construct and stuff but
it's pretty simple to understand we if you take a chart like this and slice in the big then overlaid event top so that needs to have color differences and then you can the swap the negative things up to be offset or the near the town and what happens is you squish all of that information into a much smaller space without losing out
resolution due to the top here these touch aggressive just a normal area for an squished down 3 pixels the although this this steep changes in data on top are lost when you squish about 30 pixels you don't get the same story as if you took it from a horizon and squish down we shouldn't them and the reason this works is that as humans were really good at using change and colors for determining large differences in data and position for very fine differences and if you have position for large changes then there too far away to compare if you try to use color for small changes you can't really tell the differences in color is used to make change but using both position color for they're good for using the square Streisand chart then you can get much more information
on the screen without sacrificing that sacrificing the actual quality of data this is just a random look like sinusoidal degeneration but it shows that you can really you spend a lot of time looking at this and not just trying to track down which means that you can absorb the entire picture once and that's about the thing that I'll show later on the start using right terms but explain what the theory of them 1st right so post red was
conference what's in a certain more more more more practical my favorite things is p-set state says is that as a 1 I use p-set segments from 91 with about half of so for those of you haven't p-set since the around I think since for initially by talking cure to docking and but in Postbus 9 to others just recently released September Peter gagan did some work to do great work to normalize the queries and so before it would be the sentence would show you all the queries here but if you did not I
like this is an example of the OpenNI differently just that statements before if you didn't sub-clip versions were cut by equals 1 that would show a different road and the baby to meet with 3 was 4 but it's much more useful if things are clockwise and 1 query because they have in the same group just different constants you can see there's a come more the overload of emitted but you can do with you know it's really know the broad skills useful is the total running time for a query because the city and natural readily what's written and what's of cash from anywhere dirty and so forth and so you do a fun thing with which using old had a little fun to look at the total time like queries are taken binned them up into bins and making history and with the new
feature in a 9 3 which I'm most excited about backslash watched him the sequel and you can see sort of a life history and of your queries in like these ones are taken 0 50 ms 50 to 100 100 you can see you know what this if you if you think something's wrong can see like 0 what
Microsoft take much much longer than usual others
about you know there's a bunch of was was tables that show a lot of good turtles all you know very well documented in the in the documentation was 1 of the more in parts of product because just going over the process documentation of tables and think was an interesting so you can you know the stables go through the activation of a lot of great information you know how many transactions you doing per time it fullbacks appeared in the loaded up tables of indices capturing some of the things to look at about 1 thing I did it
again with watch it is to you know cup plot a bunch these metrics sticking together with the union also and that is watch that happened and I on purpose locked a couple of my tables to see what would happen if they would strip my connections to blow up flexion state but progress a
collection of these if you if you're looking for example queries to do a lot of these introspection things this tool was made for if you're using a request this database review just go through the source code all it's actually doing is doing introspection queries and running them with peace equal to just go to this project and look at the source code you can see a good number of example queries of how do you get this sort of introspection in today's OK so you have all the information you wanted do next
on design sort of visualization system and that's unfortunate it's
not easy there's not a have and the some of the research on like visualization and I forget about because there's it's just all across the board is not confined real conceptions of that we had this is exactly what you should do and this isn't and so really have to build something and then see how it feels and iterate from there and I their that communal tough everyone's engineers began still some designers time to like help help you have you can sort of usually what I do is I see something that I like a copy of but 1 of the things
that also makes it hard is that data in and of itself is information but if you just take a metric and shove it into graphite or the broader that probably isn't enough for 2 reasons 1 other people have the same context of what this means that you do because you found it put it there and you can train them but you just having it 2nd the self explanatory the other thing is the metric in and of itself probably doesn't work you're going to need to do some sort of conversion correlated with some other metric error due to the derivatives of time you want to see how this is how this much is changing since the last on another
thing is it's really easy especially with like the graphite to just have too much stuff going on this is bad for a couple reasons 1 the it's the conceived of the green it's released by the they have like 5 6 different colors going on you can't really pieced together what's going on except OK located spikes and 1 of the things that are on the Berkeley site is that Berkeley and then another paper that says that you are slants for a charge should be at work is close to 45 degrees as you can and so that the reason that this is bad is the aspect ratio is way off because the Sun's almost vertical and if you instead made this much longer so that you can see the increases and decreases if you hear increase decreases such that it's 45 degrees then it would be much better and I found this and
this example about about searching for example about a graph or search for example graphic example graphic graph and like went through Michael terrible they are there in another thing is that the stack ones are bad because of changes in the bottom of the stack pushes everything else seek we see the parliament fine but the you know community of intuitive changes to make the top the top and they're like worthless and this also has the same problem for model and as this they weren't using yet to bounce back and forth and switch context to see any sort of correlation between these sorts of and so just brief
thing about picking colors for your visualizations I found a good PDF of color rules of the neural and I but in short which you want to do is you background should be greater muted so that colors can down to compete with that you want when you're picking colors when used like sort of softer colors that you know maybe would be found in nature but we evolved to you in nature so those colors are you have pop out a little more for similar categories you should pick 1 you vary saturation but if you contribute distinct categories you want to have different colors but it's better if you try and keep the same situation so that 1 color isn't like dominating your attention for the other ones for no reason that you at the back of the trying to the same situation that changed you then all the colors can only go further by building that makes designing
sort of visualization system part is that everyone has different and intense and different needs so working soon and visualization tool for my post as customers but many of you in this room probably deviations or contractors and each have different needs deviate you have you know a small unit to medium number of clusters databases they you know intuitively that you know very well you would have a different set like you you're actually trying really hard for performance and so the sum of these other things would be more valuable to you that a contractor who sees you know lots of databases over a long period of time and you want get in there quick and see what's going on but you don't have like a deep intuitive sense of the system and sort of taken to the extreme I'm trying to build a visualization thing for lots and lots lots of customers at Oliver's what I will probably never see the visualization that they will see it in there and that's not necessarily you know all posts experts they just wanna know why is my site slow or is it is it my fault is it your fault you and upgrade effects might have in answering those questions is different than the that other things
I said generating all this data is is tricky but there is no skills catalogs achieves all those tables but a really great to talk to to appear as a transport history unrewarding works but that that was all those variables a lot of data in that that can be pulled out and in store for visualization later I
was similar to that to that but I was I was interesting is better work you we provide a lot stream for all of your and the applications and database and this 12 factor . net site sort of like a manifest of how we think applications should be made in 1 of those factors is that blogs are a stream not files just having a bunch of log files you like process later is you that doesn't doesn't work forever instead if you treat your logs as a stream of data that can process over time it maybe story it maybe do 1 thing to do is stored but in a more importantly you can still process logs of time and notice things like that in the summer of great papers on processing you know on infinite streams of data giving stuff after getting means 4 things are so the people from the boundary I a space-saving that this disorganized porosity can bound do statistical bound on how much you're willing to throw away for counting things so you won't take infinite memory but you get accurate counts and the ones that are actually know exactly how much error there the other thing that makes
building some assistance trickier storage Ronald artistry also wanted to some great things how you can store in a lot of time series data and in doing that properties talk in and of itself and there's a lot of things you can consider on doing roll up the data over time so you can diminish space and doing the work in the retention strategies I saw really excited about the PG was that you were tation Manager extension that some of the latent talking yesterday only 1 can do that yeah yeah the boats Preussen so I had I had not heard of that before so I really want to check out because they were building up you know some interpretations of having that you've done for me if I begin easily logo really way overboard storage 1 of the things that I think I hear a lot of overlap but they take storing all that serious all the time like extremely seriously like the store metrical them and they replicated after like you know across a continent goes crazy replication stuff but honestly if I lose their 5 minutes of data in the middle like probably find like it would suck if there's a problem and that happened to be in that same 5 minutes but it probably won't be so think of Europe your realizing that sort of trade off and then designing for that is interesting it typically everyone store everything perfectly never lose anything but when you start generating so much data problems so so that I wanna go under
example of the tool that you can use a press 9 2 + database are called data scope
and this time due to this this uses that horizon check and training library of dimension is called the cubism and it's both on top of the 3 and so this is that this is an example of looking at all my collection counts application rate and some other metrics that were easy to just throw up in for a quick demo and so I had my app and scaled up at 8 PM intimate connection can you 1 way than all the workers were doing a bunch of steps you can see at that same time you can see the selects jump up how long the taken jumps up because of some contention the inserts go where high and then you can see that again I purposely lock the tables so you couldn't do anything and you can see what effect that had and everything else because all these are stacked on top of each other you can see how everything all those events are correlated and I and I felt so
bad doing topic no could also that's a codon so this is the way that this works is that there's a worker with that is correct all the time to to databases 1 1 person during in the 2nd word storing this data and it doesn't observations collects and some information from PG stat statements and some of the union catalog tables it resets that statements and sweets period again of publishing using the statistics all the time but it wasn't that it's just the piece that statement statistics and these do you said I couldn't quite calculated of this and so it did solid information stores it is Jason with the timestamp and J. Sun and then it uses a cost of expose a custom cubism out of the box Q cubism can connect to your graphite or 1 of the sources that use or you can make your own data source to give it any sort of asymmetric you want
this schema on the storage thing is reasonable I just index on created at in an IDE Jason data whatever because as I was building this new the great things about the trees and the papers I was in figuring out what things I need to store and you know the data from statements here has a lot of nasty things like for each query we store all this information so that it is a good thing I wasn't doing queries on the inside of the workers super simple if you build sophistries on hash and stores into the database and that in the presence the this is an example of how it's clicking the piece that statements they're doing other things that are from other users and then for each query getting up the query name the causes of should also throw in registered kept and fit on the slide that stores the blocks right and done so that certain open can you take a look at a point a database and see if you notice any patterns over time and we work more on that aren't getting more detailed information you can drill down in the all the data is there to do it we can drop down on individual query and see how that was performed over time I and
Streaming <Kommunikationstechnik>
Scheduling
Programmiergerät
Bit
Mereologie
Speicher <Informatik>
Code
Quick-Sort
Computeranimation
Überlagerung <Mathematik>
Eins
Metropolitan area network
Punkt
Menge
Kraft
Zahlenbereich
Quick-Sort
Computeranimation
Lineare Abbildung
Nichtlinearer Operator
Statistik
Punkt
Ausgleichsrechnung
Quick-Sort
Computeranimation
Arithmetisches Mittel
Metropolitan area network
System F
Forcing
Menge
Dezimalsystem
Korrelationsfunktion
Varianz
Logdatei
Ortsoperator
Graph
Stichprobenumfang
Mereologie
Besprechung/Interview
Information
Physikalisches System
Computeranimation
Homepage
Komplexes System
Besprechung/Interview
Interaktives Fernsehen
Visuelles System
Physikalisches System
Verkehrsinformation
Computeranimation
Homepage
Metropolitan area network
Große Vereinheitlichung
Vorzeichen <Mathematik>
Dreiecksfreier Graph
Güte der Anpassung
Visualisierung
Kanalkapazität
Vorlesung/Konferenz
Physikalisches System
Frequenz
Große Vereinheitlichung
Computeranimation
Web Site
Graph
Rahmenproblem
Güte der Anpassung
Besprechung/Interview
Summengleichung
Ungerichteter Graph
Computeranimation
Netzwerktopologie
Summengleichung
Metropolitan area network
Prognoseverfahren
Rechter Winkel
Visualisierung
Wort <Informatik>
Kantenfärbung
Gerade
Metropolitan area network
Weg <Topologie>
Zeitreihenanalyse
Linienelement
Mereologie
Datentyp
Gamecontroller
Normalvektor
Gerade
Computeranimation
Einfach zusammenhängender Raum
Kreisfläche
Graph
Zahlenbereich
Kartesische Koordinaten
Extrempunkt
Kreisdiagramm
Computeranimation
Arithmetisches Mittel
Metropolitan area network
Vorzeichen <Mathematik>
Bus <Informatik>
Gerade
Schreib-Lese-Kopf
Chirurgie <Mathematik>
Dienst <Informatik>
Ortsoperator
Hypermedia
Quellcode
Physikalische Theorie
Gerade
Metropolitan area network
Wasserdampftafel
Mereologie
Analysis
Metropolitan area network
Punkt
Graph
Wasserdampftafel
Relativitätstheorie
Cliquenweite
Quick-Sort
Gerade
Computeranimation
Dichte <Physik>
Office-Paket
Umsetzung <Informatik>
Punkt
Graph
Gefrieren
Verzweigendes Programm
Quellcode
Quick-Sort
Computeranimation
Office-Paket
Mapping <Computergraphik>
Metropolitan area network
Weg <Topologie>
Einheit <Mathematik>
Flächeninhalt
Minimum
Mereologie
Kontrollstruktur
Große Vereinheitlichung
Gerade
Verdünnung <Bildverarbeitung>
Expertensystem
Quader
Güte der Anpassung
Extrempunkt
Visuelles System
Computeranimation
Eins
Dichte <Physik>
Mapping <Computergraphik>
Visualisierung
Kantenfärbung
Information
Bandmatrix
Normalspannung
Umwandlungsenthalpie
Gerichteter Graph
Rahmenproblem
Graph
Güte der Anpassung
Gruppenoperation
Visualisierung
Ungerichteter Graph
Information
Zeitzone
Computeranimation
Eins
Web Site
Subtraktion
Punkt
Graph
Extrempunkt
Indexberechnung
Geräusch
Zeiger <Informatik>
Transinformation
Computeranimation
Metropolitan area network
Multiplikation
Einheit <Mathematik>
Automatische Indexierung
Mereologie
Visualisierung
Ablöseblase
Server
Information
Widerspruchsfreiheit
Gerade
Analysis
Folge <Mathematik>
Einfügungsdämpfung
Graph
Linienelement
Physikalischer Effekt
Datenhaltung
Relativitätstheorie
Familie <Mathematik>
Dialekt
Quick-Sort
Computeranimation
Metropolitan area network
Last
Negative Zahl
Dienst <Informatik>
Prozess <Informatik>
Automatische Indexierung
Notebook-Computer
Datentyp
Projektive Ebene
Beamer
Normalspannung
Gerade
Neumann-Problem
Softwaretest
Graph
Program Slicing
Selbstrepräsentation
Gruppenoperation
Zahlenbereich
Physikalisches System
Ereignishorizont
Quick-Sort
Raum-Zeit
Computeranimation
Metropolitan area network
Benutzerbeteiligung
Mittelwert
Information
Response-Zeit
Gerade
Maschinenschreiben
Subtraktion
Pixel
Ortsoperator
Mathematisierung
Extrempunkt
Term
Physikalische Theorie
Computeranimation
Metropolitan area network
Quadratzahl
Flächeninhalt
Rechter Winkel
Randomisierung
Kantenfärbung
Information
Touchscreen
Bildauflösung
Data Encryption Standard
Overloading <Informatik>
Total <Mathematik>
Gruppenkeim
Versionsverwaltung
Abfrage
Rechenzeit
Befehl <Informatik>
Systemaufruf
Computeranimation
Inverser Limes
Konstante
Binärdaten
Total <Mathematik>
Zählen
Statistische Analyse
Ordnung <Mathematik>
Aggregatzustand
Stabilitätstheorie <Logik>
Prozess <Physik>
Existenzaussage
Güte der Anpassung
Fortsetzung <Mathematik>
Biprodukt
Quick-Sort
Computeranimation
Eins
Metropolitan area network
Transaktionsverwaltung
Turtle <Informatik>
Mereologie
Statistische Analyse
Flächeninhalt
Indexberechnung
Information
Tabelle <Informatik>
Caching
Einfach zusammenhängender Raum
Automatische Indexierung
Linienelement
Datenhaltung
Güte der Anpassung
Indexberechnung
Abfrage
Zahlenbereich
Plot <Graphische Darstellung>
Quellcode
Quick-Sort
Computeranimation
Metropolitan area network
Arithmetische Folge
Datenerfassung
Information
Aggregatzustand
Tabelle <Informatik>
Besprechung/Interview
Visualisierung
Visuelles System
Hilfesystem
Whiteboard
Quick-Sort
Computeranimation
Web Site
Umsetzung <Informatik>
Minimalgrad
Derivation <Algebra>
Kantenfärbung
Kontextbezogenes System
Green-IT
Quick-Sort
Fehlermeldung
Subtraktion
Graph
Kategorie <Mathematik>
Natürliche Zahl
Güte der Anpassung
Mathematisierung
Dichte <Stochastik>
Schlussregel
Kontextbezogenes System
Quick-Sort
Computeranimation
Eins
Informationsmodellierung
Minimum
Visualisierung
Vorlesung/Konferenz
Kantenfärbung
Chi-Quadrat-Verteilung
Korrelationsfunktion
Soundverarbeitung
Expertensystem
Gewichtete Summe
Datenhaltung
Zahlenbereich
Online-Katalog
Physikalisches System
Frequenz
Quick-Sort
Computeranimation
Variable
Einheit <Mathematik>
Menge
Mereologie
Visualisierung
Visuelles System
Speicher <Informatik>
Cluster <Rechnernetz>
Tabelle <Informatik>
Standardabweichung
Web Site
Gewicht <Mathematik>
Prozess <Physik>
Web log
Besprechung/Interview
Kartesische Koordinaten
Zählen
Login
Raum-Zeit
Computeranimation
Eins
Streaming <Kommunikationstechnik>
Datenmanagement
Zeitreihenanalyse
Datenreplikation
Notepad-Computer
Speicher <Informatik>
Maßerweiterung
Gleitendes Mittel
Interpretierer
Kategorie <Mathematik>
Datenlogger
Logarithmus
Datenhaltung
Teilbarkeit
Quick-Sort
Unendlichkeit
Arithmetisches Mittel
Randwert
Festspeicher
Strategisches Spiel
Bitrate
Fehlermeldung
Einfach zusammenhängender Raum
Soundverarbeitung
App <Programm>
Einfügungsdämpfung
Wellenpaket
Linienelement
Hausdorff-Dimension
Datenhaltung
Kartesische Koordinaten
Extrempunkt
Bitrate
Speicherbereichsnetzwerk
Maskierung <Informatik>
Ereignishorizont
Computeranimation
Metropolitan area network
Wechselsprung
Programmbibliothek
Inhalt <Mathematik>
Horizontale
Tabelle <Informatik>
Stereometrie
Punkt
Quader
Befehl <Informatik>
Online-Katalog
Computeranimation
Netzwerktopologie
Mustersprache
Hash-Algorithmus
Statistische Analyse
Vererbungshierarchie
Luenberger-Beobachter
Zeitstempel
Speicher <Informatik>
Caching
Tabelle <Informatik>
Statistik
Befehl <Informatik>
Physikalischer Effekt
Datenhaltung
Indexberechnung
Abfrage
Quellcode
p-Block
Frequenz
Systemaufruf
Rechenschieber
Automatische Indexierung
Wort <Informatik>
Information
Programmierumgebung
Tabelle <Informatik>
Computeranimation

Metadaten

Formale Metadaten

Titel Visualizing Postgres in realtime
Serientitel PGCon 2013
Anzahl der Teile 25
Autor Leinweber, Will
Mitwirkende Heroku (Sponsor)
Lizenz CC-Namensnennung - keine kommerzielle Nutzung - Weitergabe unter gleichen Bedingungen 3.0 Unported:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen und nicht-kommerziellen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen und das Werk bzw. diesen Inhalt auch in veränderter Form nur unter den Bedingungen dieser Lizenz weitergeben
DOI 10.5446/19067
Herausgeber PGCon - PostgreSQL Conference for Users and Developers, Andrea Ross
Erscheinungsjahr 2013
Sprache Englisch
Produktionsort Ottawa, Canada

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract Postgres comes with several introspection tools out of the box. Some are easier to understand than others, but they are all useful. Recent improvements in 9.2's pgstatstatements make it even easier gain insights into the performance of your application. This talk will explore these built in tools, and what it takes to combine them to provide real-time visualizations of your database. Other topics will incldue - What metrics are the most valuable and how to use them - A deep dive into example application for realtime postgres visibility - Storage of postgres statistics - Tools for time series visualization - Collecting metrics at scale - What works and, just as importantly, what does not work

Ähnliche Filme

Loading...