Scaling Big While Sleeping Well

Scaling Big while Sleeping WellJosh Holmes@joshholmesjosh.holmes@microsoft.comwww.joshholmes.com

AgendaTesting the WatersWhat is cloud computing?What are the advantages of cloud computing?How’s Azure fit in?Diving DeepDigging into PHP/Ruby on AzureLeveraging FastCGILeveraging Azure StorageLeveraging SQL Azure

We are all excited about the cloud

Cloud-Addressable Challenges115Maintaining security while increasing access and transparency—within and outside the organizationInfrastructure costs are fixed and ongoing, and distract from the mission of reducing business process friction26Leveraging past investments to provide future valueSecurity, access, and transparency across the value chain: suppliers, partners, etc.3Many data centers at limit—real estate, capacity, cooling and power7Lack of a common platform4Matching capacity to demand

In House or Hosted ServersAllocated IT-capacitiesLoad Forecast“Under-supply“ of capacities“Waste“ of capacitiesFixed cost of IT-capacitiesIT CAPACITYBarrier forinnovationsActualLoadTIME

Cloud ComputingLoad ForecastAllocated IT capacitiesNo “under-supply“IT CAPACITYReduction of “over-supply“Possible reduction of IT-capacities in case of reduced loadReduction of initial investmentsActualLoadTime

Types of CloudsPlatform(as a Service)Private(On-Premise)Infrastructure(as a Service)You manageApplicationsApplicationsApplicationsYou manageRuntimesRuntimesRuntimesSecurity & IntegrationSecurity & IntegrationSecurity & IntegrationManaged by vendorDatabasesDatabasesDatabasesYou manageServersServersServersManaged by vendorVirtualizationVirtualizationVirtualizationServer HWServer HWServer HWStorageStorageStorageNetworkingNetworkingNetworking

The Microsoft Cloud~100 Globally Distributed Data CentersQuincy, WAChicago, ILSan Antonio, TXDublin, IrelandGeneration 4 DCs

Scaling Big While Sleeping Well

Categories of ServicesApplication ServicesSoftware ServicesPlatform ServicesThe Microsoft CloudInfrastructure Services

The trick is knowing what your systems need,and how the cloud can help.

Not everything has to or can live in the cloud

Only move to the cloud that which can benefit

The core strategy is ‘hybrid’

We call this Software + Services

“Packaged”ApplicationBig Pharmaceutical ExampleHosted “Packaged”“Packaged”using cloud “Software as a Service”BuyERP“Too costly to run this myself, but I’ve made too many customizations”CRMEmailBuild vs. Buy“Home Built”ApplicationHosted “Home Built”“Home Built”using cloud “Platform as a Service”HR SystemMolecule ResearchBuildClinical TrialHosterVendorOn PremiseCloud

“Packaged”ApplicationBig Pharmaceutical ExampleHosted “Packaged”“Packaged”using cloud “Software as a Service”BuyERPCRM“CRM and Email are commodity services – They have no customizations, and it’s cheaper for someone else to run these”EmailBuild vs. Buy“Home Built”ApplicationHosted “Home Built”“Home Built”using cloud “Platform as a Service”HR SystemMolecule ResearchBuildClinical TrialHosterVendorOn PremiseCloud

Big Pharmaceutical Example“Packaged”ApplicationHosted “Packaged”“Packaged”using cloud “Software as a Service”BuyERPCRMEmailBuild vs. Buy“Home Built”ApplicationHosted “Home Built”“Home Built”using cloud “Platform as a Service”“I can’t afford to maintain this old HR application written in VB4 – it’s driving me mad!”HR System“…but due to regulatory issues, I cannot store my HR data off-premise”Molecule ResearchBuildClinical TrialHosterVendorOn PremiseCloud

“Packaged”ApplicationBig Pharmaceutical ExampleHosted “Packaged”“Packaged”using cloud “Software as a Service”BuyERPCRMEmailHR SystemBuild vs. Buy“Home Built”ApplicationHosted “Home Built”“Home Built”using cloud “Platform as a Service”“I wish I had access to cheaper compute and storage when I need it”Molecule ResearchBuildClinical TrialHosterVendorOn PremiseCloud

“Packaged”ApplicationBig Pharmaceutical ExampleHosted “Packaged”“Packaged”using cloud “Software as a Service”BuyERPCRMEmailHR SystemBuild vs. Buy“Home Built”ApplicationHosted “Home Built”“Home Built”using cloud “Platform as a Service”Molecule ResearchBuild“THIS is where I want to spend my IT resources – I’m going to double down on this application!”Clinical TrialHosterVendorOn PremiseCloud

Only 2 reasons to use the cloud

Windows Azure VMsSmallMediumLarge X Large$0.12 $0.24 $0.48 $0.96 Per service hourPer service hourPer service hourPer service hour1 x 1.6Ghz 2 x 1.6Ghz 4 x 1.6Ghz 8 x 1.6Ghz(moderate IO) (high IO)(high IO) (high IO)1.75 GB memory 3.5 GB memory 7.0 GB memory14 GB memory

Windows Azure, In One PictureBusiness PortalDeveloper PortalService Management ServiceRESTUser Code……Runtime APIStorage ClusterVS ToolsCompute Cluster…WA SDKVM SetupRESTMSFT DatacentersDesktopCloud VM

Determine an ROIConsider:Current hosting costs

LicensingAzureROI.CloudApp.net

Digging into Azure with PHP/Ruby

Windows Azure RolesWorkerRoleInternal Facing

Think Service or DeamonWebRoleExternal Facing

Think Web ServerRunning PHP in Windows AzureHow to Do ItHost in Web role (like .NET)Supply PHP runtimePoint to runtime via FastCGI configuration inWeb.configWeb.roleConfigPHP Web RoleInstance 1VIPLoad BalancerPHP Web RoleInstance 2

Windows Azure 4 EclipseDoes a lot of the heavy lifting for youCreates the Web.config, Web.roleConfig, PHP implementation and debugs in the dev fabrichttp://windowsazure4e.org

Running EXEs in the Worker ProcessStart with a Worker Role Process to Copy the exes to the worker role’s app directoryExecute the processMonitor the health of the processOther auxiliary tasks…

Running Ruby in the Worker Processhttp://code.msdn.microsoft.com/railsonazureDownload and extract the solution Copy the Ruby bin and lib folders to the Ruby folder in the RR project Copy your rails application to the RailsApp folder in the RR project Open the solution in Visual Studio Run in the Dev FabricNavigate to http://127.0.0.1:5100

Storage optionsWindows Azure DataSql Azure

Windows Azure Data StorageQueueBlobAccountTablesDrives

Azure Platform Data Storage OptionsWindows Azure Data Storage

Semi-structured or tabular data storage

Buffered delivery data storage

Durable NTFS volumes that Windows Azure applications can use. See: http://microsoftpdc.com/Sessions/SVC14SQL AzureRelational data storageAdmin Windows Azure Data Storage

PHP with Windows Azure StorageWindows Azure SDK for PHP @ http://phpazure.codeplex.comPHP programming model for Windows Azure StorageFeatures PHP classes for Blobs, Tables & QueuesStore PHP sessions in Table Storage

Windows Azure Data Storage - BlobsUnstructured data storagePartitioned by containerUnlimited containers

Blobs Sample$blobStorage= new Microsoft_WindowsAzure_Storage_Blob();// Createif (!$blobStorage->containerExists($containerName)){$blobStorage->createContainer($containerName);$blobStorage->setContainerAcl($containerName, Microsoft_WindowsAzure_Storage_Blob::ACL_PUBLIC);}// Store$blob = $blobStorage->putBlob($containerName, $blobName, $localFilename, $metadata);/* @var $blob Microsoft_WindowsAzure_Storage_BlobInstance */

Blobs Sample Cont…// Copy$result = $blobStorage->copyBlob( $containerName, $blobName, $containerName, $blob2Name);// Retrieve$tempStore= azure_getlocalresourcepath('tempstore');$tempPath= $tempStore. '\\' . $imageId;$blobStorage->getBlob($containerName, $blobName, $tempPath);// Delete$result = $blobStorage->deleteBlob($containerName, $blobName);$result = $blobStorage->deleteContainer($containerName);

Blob Stream Wrapper$blobStorage= new Microsoft_WindowsAzure_Storage_Blob();// Register:$blobStorage->registerStreamWrapper(); // registers azure://// or$blobStorage->registerStreamWrapper('blob://'); // use blob://// Use$fp= fopen('azure://mycontainer/myfile.txt', 'r');// ...fclose($fp);

Storage: XDriveNTFS drive in the cloud: X:\Fixed size between 16MB and 1TBMax 8 drives on your VMImplemented as a page blobSingle instance writeMultiple instance read

Queue Workflow ConceptsWindows Azure Queue ProvidesGuarantee delivery (two-step consumption)Worker Dequeues Message and mark it as InvisibleWorker Deletes Message when finished processing itIf Worker role crashes, message becomes visible for another Worker to processDoesn’t guarantee “only once” deliveryDoesn’t guarantee orderingBest effort FIFOWorker RoleWeb RoleInput Queue (Work Items)Worker RoleAzure QueueWeb RoleWorker RoleWeb RoleWorker Role

Azure QueuesRemoveMessageGetMessage (Timeout)Worker RolePutMessageQueueMsg 1Msg 2Msg 2Msg 1Web RoleWorker RoleWorker RoleMsg 3Msg 4Msg 2

Loosely Coupled Work with QueuesWorker-Queue ModelLoad work in a queueMany workers consume the queueInput Queue (Work Items)Azure QueueWorker RoleWeb RoleWorker RoleWeb RoleWorker RoleWeb RoleWorker Role

Queues$queueClient= new Microsoft_WindowsAzure_Storage_Queue();// Create$result = $queueClient->createQueue('imageQueue');// Delete$queueClient->deleteQueue('imageQueue');// Add message$queueClient->putMessage('imageQueue', $message, $ttl);// Retrieve Messages$messages = $queueClient->getMessages('imageQueue', 10);foreach($messages as $message) {// Do work here...$queueClient->deleteMessage('imageQueue', $message);}

Windows Azure Data Storage - TablesSemi-Structured dataTables contain entitiesEntities contain propertiesMay be partitioned across thousands of servers.Support ACID transactions over single entitiesQueries over entire table.NET and REST interfaces

Windows Azure Data Storage – Tables (Terms Part 1)TableContains a set of entities. Entity (Row) Basic data items stored in a table. Property (Column)Single value in an entity. RowKeyUnique ID of the entity within a partitionTimestampTime it was created

Windows Azure Data Storage – Tables (Terms Part 2)Partition Entities in a table with the same partition keyPartitionKeySegments entities in to partitions to automatically distribute the table’s entities over many storage nodes.Sort OrderThere is a single index provided for the CTP, where all entities in a table are sorted by PartitionKey and then RowKey

Key Example – Blog PostsPartition 1Partition 2Getting all of dunnry’s blog posts is fastSingle partitionGetting all posts after 2008-03-27 is slowTraverse all partitions

Table Sample$tableStorage= new Microsoft_WindowsAzure_Storage_Table('table.core.windows.net', 'myaccount', 'myauthkey');// Create$result = $tableStorage->createTable($tableName);// List $result = $tableStorage->listTables();foreach($result as $table) {echo 'Table name is: ' . $table->Name . "\n";}// Delete$tableStorage->deleteTable($tableName);

Tables with Entities// Structured entityclass ImageEntityextends Microsoft_WindowsAzure_Storage_TableEntity{/** * @azure filename */public $filename;/** * @azure size Edm.Int64 */public $size;}// Unstructured entity// Microsoft_WindowsAzure_Storage_DynamicTableEntity

Tables with Entities Cont…// Insert$image = new ImageEntity($partitionKey, $rowKey);$image->filename = $_FILES['imageUpload']['name'];$image->size = $_FILES['imageUpload']['size'];$image = $tableStorageClient->insertEntity($tableName, $image);// Retrieve$image = $tableStorage->retrieveEntityById($tableName, $partitionKey, $rowKey, 'ImageEntity');// Update$image->filename = 'newname.jpg';$result = $tableStorage->updateEntity($tableName, $image);// Delete$result = $tableStorage->deleteEntity($tableName, $image);

Table Queries// Filter condition$select = "filesizegt 1024 and PartitionKeyeq '$partitionKey'";// Or fluent interface$select = $tableStorage->select()->from($tableName) ->where('filesizegt 1024') ->andWhere('PartitionKeyeq ?', $partitionKey);// Run query$images = $tableStorage->storageClient->retrieveEntities('testtable', $select, 'ImageEntity');foreach($images as $image) {echo 'Name: ' . $image->filename . "\n";}

Batching Operations// Start batch$batch = $tableStorage->startBatch();// Insert multiple$images = generateEntities();foreach($images as $image) {$tableStorage->insertEntity($tableName, $image);}// Commit$batch->commit();

Table session handler$sessionHandler = new Microsoft_WindowsAzure_SessionHandler($tableStorage);$sessionHandler->register();session_start();if (!isset($_SESSION['start'])) {$_SESSION['start'] = time();}

SQL Azure and Windows Azure Table ComparisonWindows Azure TablesSQL Azure TablesSemi-structuredLoosely typedNon-Relational (Not RDMS)Massively scalableFully structuredStrongly typedRelational (RDMS)Highly scalable

MySQL: Simple ConfigurationVIPLoad BalancerWeb RoleMySQLWorker Role

MySQL in a Windows Azure ApplicationRunning MySQL in a worker roleCopy MySQL to the worker role sub-directoryCopy to read-write local storageConfigure MySQL to listen on the right portMonitor MySQL healthConsuming MySQLDiscover IP address and portNormal access from then onHandle topology changes

Simple ConfigurationVIPLoad BalancerMySQL

ReplicationVIPLoad BalancerMSSMySQLMySQLMySQL

Windows Azure Drive with Hot SpareVIPLoad BalancerMySQLMySQL

SQL Azure FeaturesSupportedTables, Indexes, ViewsStored ProceduresTriggersConstraintsTable VariablesTemp Tables (#Name)Not SupportedPhysical Server Access Catalog DDLCommon Language RuntimeService BrokerReporting ServicesAnalysis ServicesDistributed Transactions and Queries

SQL AzureDeploymentWeb Portal(API)DB ScriptSQL AzureTDS

Scaling Big While Sleeping Well

More Related Content

What's hot (20)

Viewers also liked (7)

Similar to Scaling Big While Sleeping Well (20)

More from Josh Holmes (17)

Recently uploaded (20)

Scaling Big While Sleeping Well

Editor's Notes