New in Google Cloud: SDK 1.5.5, Cloud SQL, Cloud Storage, Prediction API and Premier Accounts
Google Cloud Services has announced a new version of the App Engine SDK (1.5.5) -the frontend request deadline has been raised from 30 sec. to 60 sec., Python 2.7-, Cloud SQL, Cloud Storage and Prediction API are out of Code Labs, and Premier Accounts.
Some of the new features coming with App Engine SDK 1.5.5 are:
- The Frontend request deadline has been raised from 30 seconds to 60 seconds.
- The number of files that can be uploaded has been raised from 3,000 to 10,000.
- The file size limit for a single file uploaded to App Engine has been raised from 10MB to 32MB.
- The online URLFetch maximum deadline has been increased from 10 seconds to 60 seconds.
- The URLFetch Post payload has been increased from 1MB to 5MB.
The SDK now includes the experimental Python 2.7 runtime for applications using the datastore. The most interesting feature of this version of the runtime is the inclusion of the webapp2 framework, coming with better URI routing and exception handling, a full featured response object and a more flexible dispatching mechanism, plus a number of optional utilities for sessions, localization, internationalization, domain and subdomain routing, secure cookies and others.
The SDK Java and SDK Python pages contain more details on this release of the App Engine SDK. The official roadmap (which seems to be a bit out of date) includes a number of features that Google aims to be implemented in the following six months:
- SSL access on custom domains
- Full-text search over Datastore
- Support for running MapReduce jobs across App Engine datasets
- Bulk Datastore Import and Export tool
- Improved monitoring and alerting of application serving
- Logging system improvements to remove limits on size and storage
Cloud SQL is a recent addition to Code Labs offering the possibility to run MySQL 5.1.59 on Google’s cloud environment. The service is free at least until the end of 2011 without a firm date when it comes out of the limited preview mode when it will no longer be free. Cloud SQL currently supports most MySQL features but not all, such as user defined functions, DB replication, LOAD_FILE(), and some statements. And the DB is limited to 10GB during preview. The database can be programmatically accessed only from Java via JDBC and Python via DB-API 2.0.
Formerly known as Google Storage for Developers, Cloud Storage has graduated from Code Labs and now can also be accessed through the App Engine Files API. Google has lowered prices at $0.13/GB/month for the first TB, lower than Amazon S3’s pricing which is $0.14/GB/month. Network transfer price associated with data is the same as Amazon’s at $0.12/GB for the first TB.
Prediction API was released in May and now has also graduated from Code Labs with two new features: support for PMML v4.01 and data anomaly detection. The API can be used to predict results based on historic data coming from “user sentiment analysis, language identification, product recommendation, message routing, gene expression, and fraud detection”. It has been used by Ford in an experiment to lower fuel consumption based on driving data.
Google has also launched Premier Accounts providing for $500/month the following: Support, a 99.95% uptime SLA, and unlimited number of applications. Support is provided only Monday through Friday from 00:00 to 18:00 PST, so there is no support during weekends.
App Engine PaaS versus Amazon WS IaaS
Google is a PaaS that has services (CloudDB, CloudStorage, High Replication Datastore, ).
Similar but different. Comparisons.
AWS RDS (MySQL or Oracle) ~= GAE CloudDB (MySQL)
AWS S3 ~= GAE CloudStorage
Simple DB (NoSQL) ~= GAE High Replication Datastore (BigTable, NoSQL)
ElasticCache (Memcached) ~= GAE CloudCache (Memcached, JCache)
AWS EC2 ~= nothing (PaaS versus IaaS mentality)
AWS Elastic Map Reduce ~= ????
??? ~= GAE Task Queue
AWS Auto Scaling ~= ???? google just handles it
AWS Spot Instances ~= ???? really a IaaS thing or could Google do something similar
AWS Cluster Compute Instances ~= ??? Nodes can't talk to each other so no need to talk to each other (except through HTTP)
AWS CloudFront (like Akami) =~ ??? they have to have this. What is it called?
??? =~ XMPP
AWS SQS (queuing) =~ ???
AWS SNS (events) =~ ???
AWS SES (email) =~ GAE Email API
AWS CloudWatch =~ ????
AWS Route53 (DNS) =~ ???
AWS Elastic Load Balancer =~ ???
AWS VPC (virtual private cloud) =~ ???
AWS Direct Connect (VLAN to connect your data center to Amazon Cloud) =~ read once that GAE has something similar, but couldn't find it.
AWS EBS (Virtual Disk that you can mount in any AWS VM instance) =~
AWS FPS (like PayPal) =~ Not provided but you can use PayPal or others
Of course if AWS does not have something Google has... you can probably find a vendor that has an Amazon VM that can provide it.
Has anyone ever done a comparison on these services for reliability, cost, ease of use, flexibility, etc.?
Has anyone compared Eclipse GAE plugins versus ElasticBean Stalk pluigins?
It seems many PaaS providers build on top of AWS. So you sort of have your choice on how you want to PaaS on AWS.
Amazon seems to have better branding. In the Google docs, things change names and are not consistently replicated throughout other docs. Google docs for GAE seems to be more flea market book store. Amazon docs are more like a public library with a good librarian.
Just some random thoughts... I hope someone corrects me and fills in some of the blanks so I learn some new stuff.
Re: App Engine PaaS versus Amazon WS IaaS
Interesting time ahead.