Search Results

Search found 9156 results on 367 pages for 'cloud storage'.

Page 71/367 | < Previous Page | 67 68 69 70 71 72 73 74 75 76 77 78  | Next Page >

  • Geek City: Where are LOBs stored?

    - by Kalen Delaney
    When researching a question from one of the students in my class last week, I was reading the documentation for CREATE TABLE about storing LOB columns at http://msdn.microsoft.com/en-us/library/ms174979.aspx . For this discussion LOB columns includes text, image, ntext, xml and the MAX columns when they are over 8000 bytes and stored outside the regular data row. I knew that SQL Server gives us the capability of storing LOB columns in a separate filegroup with the TEXTIMAGE_ON clause, but I was surprised...(read more)

    Read the article

  • Consumer Electronics Show (CES) Summit:Best Practices in Transforming Channels and Partnerships

    - by charles.knapp
    Expanding consumer demand is driving the entire high technology industry, accompanied by product lifecycles as short as a few months, continued pricing and promotion pressures, and increased globalization. Unifying global channel management, operations, and execution flow will increase efficiency and growth. IT can help, but one must think beyond generic ERP and CRM. Please join Oracle and IBM at the Bellagio Hotel in Las Vegas, Wednesday January 5, 1-7 pm. Learn from IBM, VTech, Plantronics, Cisco, Symantec and Oracle High Tech Product Strategy how to improve:Channel sales, marketing, and operations management - enhance NPI, sales, forecasts, training, promotion planning, execution and settlement Winning the deal - determining the right price for the right deal for the "perfect quote", capturing the order and order management Collaborative and rapid supply chain planning - improve agility, inventory turns, and profits Register now for this FREE event. We hope you'll join us for our Oracle High Technology CES Summit and networking reception with your peers.

    Read the article

  • Webinar: Integrated Sales & Marketing - An Impossible Dream?

    - by charles.knapp
    Are you making the most of the latest B2B marketing thinking? Are your marketing tactics, your outbound email campaigns and your SEO generating enough of the prospects and leads that your sales teams need? Are your sales and marketing functions aligned and working together with optimised results? In this Webinar with MarketingWeek Magazine, find out how: - To ensure your marketers create and deliver consistently effective, and targeted campaigns - You can triple the customer intelligence your marketers gather, ensuring your sales teams are better informed and qualified than ever before - Generate up to 200% growth in lead volume and start measuring marketing effectiveness against increase in sales and size of an average deal - And hear how BPI OnDemand has delivered integrated sales & marketing across industries, with results such as 100% ROI on system cost for Heal's after just one campaign

    Read the article

  • How to spread XML Sitemaps over several webservers behind AWS loadbalancer?

    - by Jurik
    We have a web portal with almost a million products and way more other urls. I wrote a script that checks database. If there is a new url needed or an old one update, this script will update/create the XML Sitemaps. But we have several servers behind the load balancer at our rented AWS space. Further this script checks database for each url if there was an update so that it updates the appropriate xml file too. My question is how to spread those XML Sitemaps over all webservers behind this AWS load balancer? Our approaches/ideas: we could just generate them on one server with a cron job and copy them to the other servers, but this could be difficult because of automatic raising numbers of servers and so on. we put them on our S3 - but this one is not avaible thru our domain, so I guess google will have a problem with it I let my script run on every webserver but change it in a way that it will generate each time all xml files if they do not exist. But then I would have conflicts with updated URLs in my database, where I saved timestamp of last changed value of every url Is there another - better - solution that I do not know? Are there any special services by amazon for such cases?

    Read the article

  • Storing Attendance Data in database

    - by Ali Abbas
    So i have to store daily attendance of employees of my organisation from my application . The part where I need some help is, the efficient way to store attendance data. After some research and brain storming I came up with some approaches . Could you point me out which one is the best and any unobvious ill effects of the mentioned approaches. The approaches are as follows Create a single table for whole organisation and store empid,date,presentstatus as a row for every employee everyday. Create a single table for whole organisation and store a single row for each day with a comma delimited string of empids which are absent. I will generate the string on my application. Create different tables for each department and follow the 1 method. Please share your views and do mention any other good methods

    Read the article

  • Best way to store a large amount of game objects and update the ones onscreen

    - by user3002473
    Good afternoon guys! I'm a young beginner game developer working on my first large scale game project and I've run into a situation where I'm not quite sure what the best solution may be (if there is a lone solution). The question may be vague (if anyone can think of a better title after having read the question, please edit it) or broad but I'm not quite sure what to do and I thought it would help just to discuss the problem with people more educated in the field. Before we get started, here are some of the questions I've looked at for help in the past: Best way to keep track of game objects Elegant way to simulate large amounts of entities within a game world What is the most efficient container to store dynamic game objects in? I've also read articles about different data structures commonly used in games to store game objects such as this one about slot maps, but none of them are really what I'm looking for. Also, if it helps at all I'm using Python 3 to design the game. It has to be Python 3, if I could I would use C++ or Unityscript or something else, but I'm restricted to having to use Python 3. My game will be a form of side scroller shooter game. In said game the player will traverse large rooms with large amounts of enemies and other game objects to update (think some of the larger areas in Cave Story or Iji). The player obviously can't see the entire room all at once, so there is a viewport that follows the player around and renders only a selection of the room and the game objects that it contains. This is not a foreign concept. The part that's getting me confused has to do with how certain game objects are updated. Some of them are to be updated constantly, regardless of whether or not they can be seen. Other objects however are only to be updated when they are onscreen (for example, an enemy would only be updated to react to the player when it is onscreen or when it is in a certain range of the screen). Another problem is that game objects have to be easily referable by other game objects; something that happens in the player's update() method may affect another object in the world. Collision detection in games is always a serious problem. I need a way of containing the game objects such that it minimizes the number of cases when testing for collisions against one another. The final problem is that of creating and destroying game objects. I think this problem is pretty self explanatory. To store the game objects then I've considered a number of different methods. The original method I had was to simply store all the objects in a hash table by an id. This method was simple, and decently fast as it allows all the objects to be looked up in O(1) complexity, and also allows them to be deleted fairly easily. Hash collisions would not be a major problem; I wasn't originally planning on using computer generated ids to store the game objects I was going to rely on them all using ids given to them by the game designer (such names would be strings like 'Player' or 'EnemyWeapon4'), and even if I did use computer generated ids, if I used a decent hashing algorithm then the chances of collisions would be around 1 in 4 billion. The problem with using a hash table however is that it is inefficient in checking to see what objects are in range of the viewport. Considering the fact that certain game objects move (as well as the viewport itself), the only solution I could think of in order to only update objects that are in the viewport would be to iterate through every object in the hash table and check if it is in the viewport or not, updating only the ones that are in the valid area. This would be incredibly slow in scenarios where the amount of game objects exceeds 500, or even 200. The second solution was to store everything in a 2-d list. The world is partitioned up into cells (a tilemap essentially), where each cell or tile is the same size and is square. Each cell would contain a list of the game objects that are currently occupying it (each game object would be inserted into a cell depending on the center of the object's collision mask). A 2-d list would allow me to take the top-left and bottom-right corners of the viewport and easily grab a rectangular area of the grid containing only the cells containing entities that are in valid range to be updated. This method also solves the problem of collision detection; when I take an entity I can find the cell that it is currently in, then check only against entities in it's cell and the 8 cells around it. One problem with this system however is that it prohibits easy lookup of game objects. One solution I had would be to simultaneously keep a hash table that would contain all the positions of the objects in the 2-d list indexed by the id of said object. The major problem with a 2-d list is that it would need to be rebuilt every single game frame (along with the hash table of object positions), which may be a serious detriment to game speed. Both systems have ups and downs and seem to solve some of each other's problems, however using them both together doesn't seem like the best solution either. If anyone has any thoughts, ideas, suggestions, comments, opinions or solutions on new data structures or better implementations of the existing data structures I have in mind, please post, any and all criticism and help is welcome. Thanks in advance! EDIT: Please don't close the question because it has a bad title, I'm just bad with names!

    Read the article

  • How would you gather client's data on Google App Engine without using Datastore/Backend Instances too much?

    - by ruslan
    I'm relatively new to StackExchange and not sure if it's appropriate place to ask design question. Site gives me a hint "The question you're asking appears subjective and is likely to be closed". Please let me know. Anyway.. One of the projects I'm working on is online survey engine. It's my first big commercial project on Google App Engine. I need your advice on how to collect stats and efficiently record them in DataStore without bankrupting me. Initial requirements are: After user finishes survey client sends list of pairs [ID (int) + PercentHit (double)]. This list shows how close answers of this user match predefined answers of reference answerers (which identified by IDs). I call them "target IDs". Creator of the survey wants to see aggregated % for given IDs for last hour, particular timeframe or from the beginning of the survey. Some surveys may have thousands of target/reference answerers. So I created entity public class HitsStatsDO implements Serializable { @Id transient private Long id; transient private Long version = (long) 0; transient private Long startDate; @Parent transient private Key parent; // fake parent which contains target id @Transient int targetId; private double avgPercent; private long hitCount; } But writing HitsStatsDO for each target from each user would give a lot of data. For instance I had a survey with 3000 targets which was answered by ~4 million people within one week with 300K people taking survey in first day. Even if we assume they were answering it evenly for 24 hours it would give us ~1040 writes/second. Obviously it hits concurrent writes limit of Datastore. I decided I'll collect data for one hour and save that, that's why there are avgPercent and hitCount in HitsStatsDO. GAE instances are stateless so I had to use dynamic backend instance. There I have something like this: // Contains stats for one hour private class Shard { ReadWriteLock lock = new ReentrantReadWriteLock(); Map<Integer, HitsStatsDO> map = new HashMap<Integer, HitsStatsDO>(); // Key is target ID public void saveToDatastore(); public void updateStats(Long startDate, Map<Integer, Double> hits); } and map with shard for current hour and previous hour (which doesn't stay here for long) private HashMap<Long, Shard> shards = new HashMap<Long, Shard>(); // Key is HitsStatsDO.startDate So once per hour I dump Shard for previous hour to Datastore. Plus I have class LifetimeStats which keeps Map<Integer, HitsStatsDO> in memcached where map-key is target ID. Also in my backend shutdown hook method I dump stats for unfinished hour to Datastore. There is only one major issue here - I have only ONE backend instance :) It raises following questions on which I'd like to hear your opinion: Can I do this without using backend instance ? What if one instance is not enough ? How can I split data between multiple dynamic backend instances? It hard because I don't know how many I have because Google creates new one as load increases. I know I can launch exact number of resident backend instances. But how many ? 2, 5, 10 ? What if I have no load at all for a week. Constantly running 10 backend instances is too expensive. What do I do with data from clients while backend instance is dead/restarting? Thank you very much in advance for your thoughts.

    Read the article

  • Cannot install 14.04 on Dell Inspiron 5447 14

    - by user292121
    I install Ubuntu version 14.04 by use USB stick is a boot devices on DELL Inspiron 5447 14" but it stuck in ubuntu logo. I change to use Linux Mint and it still stuck in the same process is pending in Mint logo. When I change to Linux Mint Compatibility Mode it shown some error meassage. For the stuck screen please see the url below: http://www.image-share.com/upload/2583/266.jpg For the error massage when I try to use Linux Mint can see as below: http://www.image-share.com/ijpg-2583-268.html How can i do next ?

    Read the article

  • New Oracle Database Performance and Tuning Specialization version available to all partners!

    - by Catalin Teodor
    Any partner working with the Oracle database should take advantage of the new Oracle Database Performance and Tuning specialization version. Have partners review the criteria and see what’s new in this specialization solution update. Partners should plan to take the new version as the Oracle Database 11g Performance Tuning specialization becomes non-qualifying as of August 1, 2015.

    Read the article

  • MaaS minimum requirements with juju-jitsu?

    - by Christopher Shen Mu Long
    I've browsed through so many different sites and found so much contradictory information. As I am getting tired of this and do belive this question affects many other users, so I would like to collect the "once and for all times" answer. Unfortunately, the documentation on MaaS and Juju is ... well, not the best, sorry to say that. What are the minimum system requirements for setting up a MaaS cluster, which is going to be orchestrated with juju-jitsu? Do they need to have the exact system specifications or can I just combine different hardware? What are the minimum requirements for the master machine? E.g. "You need at least 8GB of RAM, a dual core CPU with at least 3.0 GHz." How many machines to I need to deploy MaaS on? I've read six machines, nine machines, and so on. I clearly want to know: "You need one for the Master and e.g. five nodes." Do I need to attach as many NICs (network interface cards) to my master machine as there are nodes, or can I simply attach two NICs and a switch? One NIC for connecting to the internet, one for handling the MaaS tasks, connected to a switch, which connects my nodes to the master? Is Juju now ready for local deployment? The last time I experimented with juju and had to reboot my machine, the services orchestrated by juju were gone. This was an issue I also found on the official juju site. Unfortunately, as mentioned above, the documentation is not the best, so I could not find the necessary info on that again. So: Can I use juju on a local environment or will a reboot break my setup?

    Read the article

  • Does something like this "dream" online IDE product exist?

    - by Dan Tao
    I was thinking the other day, it would be amazing if a web-based product with the following features existed: Customizable text editor with nice formatting like ACE Real-time collaborative editing like Google Docs (or the late Wave) Online multi-language compiling capabilities like Ideone.com SCM hosting and/or integration + issue management like... oh, I don't know, GitHub Clearly (considering the examples), all the desired features exist. Is there anywhere that they exist all in one product? If not, does anything come close?

    Read the article

  • front usb wont mount harddrives, internal usb ports do

    - by Thesgsuser
    I have noticed something in my new build, i am using Ubuntu desktop newest version my motherboard is the asus f1a75-m pro R2.0 with the usb ports in the back all my NTFS hard disks or usb sticks work fine, but then.. when i put them in the front usb ports of my chassis (silverstone milo ml-03) they wont mount... I have 2 usb 3.0 ports in front of the case connected with a internal usb 3.0 header. But i verified that the usb 3.0 ports on the back do mount the harddisk so it has nothing to do with usb 3.0 i think. The strange thing is, my mouse works fine on the front usb ports. Every usb hardware piece seems to work except if it has any memory inside it :( What seems to be the problem?

    Read the article

  • What tools, libraries, or framework is needed to create a completely offline Javascript application?

    - by makerofthings7
    I am interested in creating a HTML application that can run as disconnected from the server as possible. Two examples of this include OWA in Exchange 2013 and to a lesser extent and the client available at www.ripple.com With the focus on OWA in Exchange 2013, what is needed to replicate the offline functionality available in a different application? A list technologies, frameworks, etc would be immensely helpful

    Read the article

  • Store HighRes photos in Database or as File?

    - by David
    I run a site which has a couple of million photos and gets over 1000 photos uploaded each day. Up to now, we haven't kept the original file that was uploaded to conserve on space. However, we are getting to a point that we are starting to see a need to have high-res original versions. I was wondering if its better to store these in the filesystem as an actual file or if its better to store them in a database (ie: mysql). The highres images would be rarely referenced but may be used when someone decides to download it or we decide to use it for rare processes like making a new set of thumbnails sizes/etc.

    Read the article

  • How are Java ByteBuffer's limit and position variable's updated?

    - by Dummy Derp
    There are two scenarios: writing and reading Writing: Whenever I write something to the ByteBuffer by calling its put(byte[]) method the position variable is incremented as: current position + size of byte[] and limit stays at the max. If, however, I put the data in a view buffer then I will have to, manually, calculate and update the position Before I call the write(ByteBuffer) method of the channel to write something, I will have to flip() the Bytebuffer so that position points to zero and limit points to the last byte that was written to the ByteBuffer. Reading: Whenever I call the read(ByteBuffer) method of a channel to read something, the position variable stays at 0 and the limit variable of the ByteBuffer points to the last byte that was read. So, if the ByteBuffer is smaller than the file being read, the limit variable is pushed to max This means that the ByteBuffer is already flipped and I can proceed to extracting the values from the ByteBuffer. Please, correct me where I am wrong :)

    Read the article

  • Where to store front-end data for "object calculator"

    - by Justin Grahn
    I recently have completed a language library that acts as a giant filter for food items, and flows a bit like this :Products -> Recipes -> MenuItems -> Meals and finally, upon submission, creates an Order. I have also completed a database structure that stores all the pertinent information to each class, and seems to fit my needs. The issue I'm having is linking the two. I imagined all of the information being local to each instance of the product, where there exists one backend user who edits and manipulates data, and multiple front end users who select their Meal(s) to create an Order. Ideally, all of the front end users would have all of this information stored locally within the library, and would update the library on startup from a database. How should I go about storing the data so that I can load it into the library every time the user opens the application? Do I package a database onboard and just load and populate every time? The only method I can currently conceive of doing this, even if I only have 500 possible Product objects, would require me to foreach the list for every Product that I need to match to a Recipe and so on and so forth every time I relaunch the program, which seems like a lot of wasteful loading. Here is a general flow of my architecture: Products: public class Product : IPortionable { public Product(string n, uint pNumber = 0) { name = n; productNumber = pNumber; } public string name { get; set; } public uint productNumber { get; set; } } Recipes: public Recipe(string n, decimal yieldAmt, Volume.Unit unit) { name = n; yield = new Volume(yieldAmt, unit); yield.ConvertUnit(); } /// <summary> /// Creates a new ingredient object /// </summary> /// <param name="n">Name</param> /// <param name="yieldAmt">Recipe Yield</param> /// <param name="unit">Unit of Yield</param> public Recipe(string n, decimal yieldAmt, Weight.Unit unit) { name = n; yield = new Weight(yieldAmt, unit); } public Recipe(Recipe r) { name = r.name; yield = r.yield; ingredients = r.ingredients; } public string name { get; set; } public IMeasure yield; public Dictionary<IPortionable, IMeasure> ingredients = new Dictionary<IPortionable,IMeasure>(); MenuItems: public abstract class MenuItem : IScalable { public static string title = null; public string name { get; set; } public decimal maxPortionSize { get; set; } public decimal minPortionSize { get; set; } public Dictionary<IPortionable, IMeasure> ingredients = new Dictionary<IPortionable, IMeasure>(); and Meal: public class Meal { public Meal(int guests) { guestCount = guests; } public int guestCount { get; private set; } //TODO: Make a new MainCourse class that holds pasta and Entree public Dictionary<string, int> counts = new Dictionary<string, int>(){ {MainCourse.title, 0}, {Side.title , 0}, {Appetizer.title, 0} }; public List<MenuItem> items = new List<MenuItem>(); The Database just stores and links each of these basic names and amounts together usings ID's (RecipeID, ProductID and MenuItemID)

    Read the article

  • which service to use among Amazon's free tier for Java app

    - by vikas devde
    this is the first time I am going to host a java app, Amazon offers a free tier which provides free usage for one year, and I am going to use it. But there are so many services (S3, EC2, etc etc), I cant figure out which service is for web hosting, which service is for what use, their docs is so huge, I am confused, what to read. can anybody write some good points, about which service to use specifically for Java apps, how much I will be charged, and they ask for credit/debit card credential for signup, does it mean they will debit me even in the free period?

    Read the article

  • How to mange big amount users at server side?

    - by Rami
    I built a social android application in which users can see other users around them by gps location. at the beginning thing went well as i had low number of users, But now that I have increasing number of users (about 1500 +100 every day) I revealed a major problem in my design. In my Google App Engine servlet I have static HashMap that holding all the users profiles objects, currenty 1500 and this number will increase as more users register. Why I'm doing it Every user that requesting for the users around him compares his gps with other users and check if they are in his 10km radius, this happens every 5 min on average. That is why I can't get the users from db every time because GAE read/write operation quota will tare me apart. The problem with this desgin is As the number of users increased the Hashmap turns to null every 4-6 hours, I thing that this time is getting shorten but I'm not sure. I'm fixing this by reloading the users from the db every time I detect that it became null, But this causes DOS to my users for 30 sec, So I'm looking for better solution. I'm guessing that it happens because the size of the hashmap, Am I right? I have been advised to use spatial database, but that mean that I can't work with GAE any more and that mean that I need to build my big server all over again and lose my existing DB. Is there something I can do with the existing tools? Thanks.

    Read the article

  • how to setup hive on a single node?

    - by Harman
    I successfully setup hadoop on ubuntu 10.04 on a single node by going through the steps mentioned in Michael Noll's tutorial ( Running Hadoop On Ubuntu Linux (Single-Node Cluster) ). Now, I'm trying to setup hive on the same machine but I'm stuck as of what to do after I decompress the hive-0.8.1-bin.tar.gz and move it to /usr/local/hive. Any help would be appreciated but as I'm new to Linux, it would be very helpful if someone could help me step-by-step.

    Read the article

  • The Best Articles for Backing Up and Syncing Your Data

    - by Lori Kaufman
    World Backup Day is March 31st and we decided to provide you with some useful information to make backing up your data easier. We’ve published articles about backing up various types of data and settings both offline and online. There’s all kinds of settings on your computer to backup in addition to your personal data, such as Wi-Fi passwords, drivers, and settings for programs like web browsers, Office, and Windows Live Writer. There are also many tools available to help you keep your data and settings backed up. Make Your Own Windows 8 Start Button with Zero Memory Usage Reader Request: How To Repair Blurry Photos HTG Explains: What Can You Find in an Email Header?

    Read the article

  • Using EC2 instance as main development platform

    - by David
    My problem I am working as a consultant for various companies. Each company provides me with a laptop where with their software on and I also have my own where I have my development environment. I tend to buy a new laptop every second year and find myself spending lots of time configuring and installing software. I also sometimes spend a lot of time waiting for my laptop to process things. To solve all these issues, I am now considering using EC2 (running windows instances) as my main development platform and just access this from any PC I happen to be at. I calculated that running the High-CPU On-Demand Instances (medium) for 8 hours a day for a year costs me 580$, which is acceptable. I imagine that when I approach the workplace each day, I will make a single click my phone to fire up the instance, so it is ready when I get to work. I should have different icons on my phone to fire up the various instance types. The same software should of course automatically be loaded on the various hardware (sometimes I would even need their instance with 68.4 GB of memory). Another advantage is that if I am having a specific problem with my instance, I could fire up another instance and have someone look into the problem and update the image. My question: Does anyone have experience with such a setup on EC2? What kind of problems do you forsee?

    Read the article

  • Using EC2 instance as main development platform

    - by David
    My problem I am working as a consultant for various companies. Each company provides me with a laptop with their software on and I also have my own, where I have my development environment. I tend to buy a new laptop every second year and find myself spending lots of time configuring and installing software. I also spend a lot of time waiting for my laptop to process things. To solve all these issues, I am now considering using EC2 (running windows instances) as my main development platform and just access this from any PC I happen to be at. I calculated that running the Large instance (cheapest 64-bit) for 8 hours a day for a year costs me 960$ per year, which is acceptable. I imagine that when I approach the workplace each day, I will make a single tap on my phone to fire up the instance, so it is ready when I get to work. I should have different icons on my phone to fire up the various instance types. The same software should of course automatically be loaded on the various hardware (sometimes I would even need their instance with 68.4 GB of memory). Another advantage is that if I am having a specific problem with my instance, I could fire up another instance and have someone look into the problem and update the image. My question: Does anyone have experience with such a setup on EC2? What kind of problems do you foresee?

    Read the article

  • jungledisk 3.16 doesn't launch

    - by Angelo
    Has anyone had success with jungledisk 3.16 on ubuntu 11.10? I installed it from the .deb file provided by jungledisk. The install goes fine, but I can't get the "jungle disk desktop" app to launch. It appears in the dash search bar, but doesn't launch or do anything upon selecting it. When I try the command line, I get the following... me@myComputer:~$ jungledisk -V -f Verbose mode enabled Shutting down... me@myComputer:~$ What's the deal here? Anybody else experience this? Does anyone have suggestions for what to try? I opened up a help-ticket with jungledisk, but they just asked me for which ubuntu version and which gui I was using and then went silent. I've used jungledisk since 2008 and had no problems. It is sad that it is not working on the new ubuntu for me. Should I just quit them and use dropbox or one? (those seem to be working)

    Read the article

  • How can I gather client's data on Google App Engine without using Datastore/Backend Instances too much?

    - by ruslan
    One of the projects I'm working on is online survey engine. It's my first big commercial project on Google App Engine. I need your advice on how to collect stats and efficiently record them in DataStore without bankrupting me. Initial requirements are: After user finishes survey client sends list of pairs [ID (int) + PercentHit (double)]. This list shows how close answers of this user match predefined answers of reference answerers (which identified by IDs). I call them "target IDs". Creator of the survey wants to see aggregated % for given IDs for last hour, particular timeframe or from the beginning of the survey. Some surveys may have thousands of target/reference answerers. So I created entity public class HitsStatsDO implements Serializable { @Id transient private Long id; transient private Long version = (long) 0; transient private Long startDate; @Parent transient private Key parent; // fake parent which contains target id @Transient int targetId; private double avgPercent; private long hitCount; } But writing HitsStatsDO for each target from each user would give a lot of data. For instance I had a survey with 3000 targets which was answered by ~4 million people within one week with 300K people taking survey in first day. Even if we assume they were answering it evenly for 24 hours it would give us ~1040 writes/second. Obviously it hits concurrent writes limit of Datastore. I decided I'll collect data for one hour and save that, that's why there are avgPercent and hitCount in HitsStatsDO. GAE instances are stateless so I had to use dynamic backend instance. There I have something like this: // Contains stats for one hour private class Shard { ReadWriteLock lock = new ReentrantReadWriteLock(); Map<Integer, HitsStatsDO> map = new HashMap<Integer, HitsStatsDO>(); // Key is target ID public void saveToDatastore(); public void updateStats(Long startDate, Map<Integer, Double> hits); } and map with shard for current hour and previous hour (which doesn't stay here for long) private HashMap<Long, Shard> shards = new HashMap<Long, Shard>(); // Key is HitsStatsDO.startDate So once per hour I dump Shard for previous hour to Datastore. Plus I have class LifetimeStats which keeps Map<Integer, HitsStatsDO> in memcached where map-key is target ID. Also in my backend shutdown hook method I dump stats for unfinished hour to Datastore. There is only one major issue here - I have only ONE backend instance :) It raises following questions on which I'd like to hear your opinion: Can I do this without using backend instance ? What if one instance is not enough ? How can I split data between multiple dynamic backend instances? It hard because I don't know how many I have because Google creates new one as load increases. I know I can launch exact number of resident backend instances. But how many ? 2, 5, 10 ? What if I have no load at all for a week. Constantly running 10 backend instances is too expensive. What do I do with data from clients while backend instance is dead/restarting?

    Read the article

< Previous Page | 67 68 69 70 71 72 73 74 75 76 77 78  | Next Page >