Comparison between Amazon S3 and Google Cloud Storage

Amazon S3, Google Cloud Storage

The aim is to identify the differences and similarities between AWS S3 and GCS as both companies are competitors in the public cloud computing market. In many ways, both platforms are offering the same services but are somehow different from each other. To differentiate both platforms we considered factors like region, zones, Accounts, limits, and pricing, Resource management interfaces, Service types, Storage classes, Cloud Storage Performance, and Migration. The purpose of this comparison was to come up with a recommendable solution. The research was conducted by using primary resources and secondary resources. As a primary resource, we sign up with both platforms and observe the differences whereas as a secondary resource we considered journals, articles, and websites. The significance of the study is that Google Cloud Storage averaged more than three times higher latency as compared to Amazon S3. And in other factors, it's more recommendable than Amazon S3.

Cloud Storage signifies virtual storage platform which can extend to various other physical storage devices. In cloud storage, some of the user's file can be on a physical server in California, and some can be on a physical server in New York. Most of the users do not know where exactly their file is, as cloud storage considered as a vague and untouchable thing [1]. In this report, we signify the difference between cloud storage providers Amazon Web Services and Google Cloud Platform.

Amazon Simple Storage Service is a web-based interface used by users to save and retrieve any size of data at any moment, and from any location using Amazon website [2]. Google Cloud Storage (GCS) is unified object storage for engineers and ventures from computing (live information) to data analytics/ML to data archiving [3].

Both services can be used by users like could administrators, software developers, and another enterprise IT professionals through dedicated network connections or public internet. Cloud Storage allow users to store unstructured objects in containers called Buckets. They can serve static data directly from Cloud Storage or can store data for other Google Cloud Platform services. The user can create a bucket by providing information like bucket name, Default storage class, Multi-Regional location, or by adding a label. Bucket name must be unique across Cloud Storage. For privacy user should not include sensitive information in bucket name. Others can discover bucket name if it matches a name they're trying to use.

The Default storage class include Multi-Regional, Regional, Nearline, and Coldline. This report elaborates the differences and similarities between AWS S3 and GCS as both companies are competitors in the public cloud computing market [4].

BACKGROUND

Amazon Web Services (AWS) is a cloud computing platform provided by Amazon.com. Web services called as remote computing services or cloud services [5]. Amazon Web Services allow customers to try it for free. Once a client signs up, it charges customer with 1$ and provides with 12 months of free tier access including the use of Amazon EC2, Amazon S3, and Amazon DynamoDB. The user will not be auto charged during the free trial, but once free trial ends user will be auto charged. There are two types of accounts; Personal account and business. AWS offers different payment methods. AWS accepts most major credit cards. Some credit cards might require additional authorization from customer's bank to be used successfully to pay AWS or AISPL bills. They accept debit cards like Visa or MasterCard. They accept prepaid cards and VCCs (Virtual Credit Cards), but there are specific restrictions applied for using those cards. Payment is processed automatically by default, so customers do not need to worry about services will stop if payments not made. AWS does not offer an option to make an advance payment. However, some individual AWS services offer payment options that are like prepayment. For example, Amazon EC2 offers Reserved Instances, which allow the user to pay an up-front fee to reserve capacity for an account in one-year or three-year terms.

AWS offers storages like Amazon S3, Amazon Glacier, AWS Storage Gateway, AWS import/export Snowball, Amazon Elastic File System, Amazon EBS, and Amazon CloudFront. After sign up, account information is asked to identify verification. Identification verification provides with three options; provide a telephone number, call in progress, and identify verification complete.

AWS support offers to provide with different plans according to user's requirement. All programs provide 24x7 access to AWS documentation, customer service, support forums, and whitepapers. Users are recommended to select support plan which aligns with AWS usage to access to technical assistance and additional resources which can help users to plan, deploy, and optimized AWS environment. The user is required to select one option from basic, developer, business, and enterprise. Basic is for free. Basic is for regular use. The developer is for users who are experimenting with AWS starts at $29/month (scales based on usage). Business is for Production use of AWS starts at $100/month (scales based on usage). Enterprise is for the mission-critical use of AWS starts at $15,000/month (scales based on usage) If user select enterprise option, customer support will contact user within 48 hours to discuss their needs and finalize the signup. Support resources will be available when subscription completed, and no charges incurred until that time.

Google Cloud Platform considered as a collection of public cloud computing services offered by Google [3]. Google Cloud Platform allow customers to try it for free. Once a client signs up, it charges customer with 1$ and provides with $300 credit for free to spend on Google Cloud Platform for over 12 months. That account allows clients to access to all Cloud Platform products like running application, websites, and services. The user will not be auto charged during, and after the free trial ends, Google ask for credit card to make sure that it's not a robot. Once the user enters their credit card details, it will monthly automatically charge from users account. There are two types of accounts; Individual and business. Google Cloud Platform offers different payment methods. They accept debit cards like Visa or MasterCard. They do not accept prepaid cards and VCCs (Virtual Credit Cards). Payment is processed automatically by default, so customers do not need to worry about services will stop if payments not made. It also allows users to make payments in advance. Google Cloud Platform offers storages like Bigtable, SQL, Datastore, Storage, and spanner. Identify verification option not found while signing up.

Google Cloud Platform support level includes Bronze, silver, gold, and platinum. Bronze is for free and provides with $300 credit for free to spend on Google Cloud Platform for over 12 months. Silver provides with lowest-cost access to support services for $150/Month. Gold is for production services on Cloud Platform which starts at $400/month. Platinum level is for high volume production services on Cloud Platform and price is given by Amazon after requirements provided.

Amazon S3 VS Google Cloud Storage

In this section, we discuss differences and similarities between Amazon S3 and Google Cloud Storage by considering distinct factors like Regions and Zones, Accounts, limits, and pricing, Resource management interfaces, and Service types.

A. Regions and Zones

Amazon S3 services are implemented within areas location globally like Asia Pacific (Mumbai, Seoul, Singapore, Sydney, Tokyo), Canada (Central), EU (Frankfurt, Ireland, London), South America (Sao Paulo), and US East (N Virginia, Ohio, N. California, Oregon). Each region comprises a group of data centers that are comparatively close to each other. Respectively, Google Cloud Storage available in multi-regional locations like the US (US-Central1, US-East1, US-West1), Asia (Asia-East1, Asia-NorthEast1), and Europe (Europe-West1)

Each region of AWS and GCS is autonomous and isolated from other AWS regions. Both Cloud Storage platform have POPs (Points of Presence) resides in different location globally, but both uses it in a different way. The table given below shows the terms and concepts to map AWS and Google Cloud Platform. AWS use POPs to deliver CDN (Content Delivery Network) service like Amazon CloudFront whereas GCS uses POPs (Points of Presence) to deliver Google Cloud CDN and built-in edge caching for services like Cloud Storage and App Engine [6].

The figure 1 given below shows the concepts of AWS and GCS.

Figure 1: Regions and Zones

To deploy the services, users are required to select a data near their number one targets of users. As an example: If users are doing retail web hosting in the West Coast of US and consumer wants to deploy their services proper there to lessen the latency and provide better user experience. Deployment can be using from afar, but User Experience will suffer [7].

The figure given below shows that Amazon has most extensive coverage in US West, US East, Canada, South America, Europe, Asia-Pacific, China. Soon it is going to cover Paris and Ningxia as well [8].

Figure 2: Amazon Coverage Regions

The figure given below shows the Google Cloud Storage coverage in the US. It falls in the back of in Europe and specifically in Asia only one spot in Taiwan without coverage in South America and Africa [9].

Figure 3: Google Coverage Regions

According to Jishnu Kinwar, Time taken to upload content to Singapore Region (SG) was higher than uploading content to North Carolina (NC) region. The solution to that problem was uploading content to AWS S3 (NC), and then copy bucket into SG region was faster than directly uploading to SG region. This solution reduces a time taken to upload content, but users need to manage two buckets and pay for two network charges (NC- SG) and (SG-CDN). To avoid managing two buckets and to pay for two network costs alternative is Google Cloud Storage as they have the better network and use dark fiber to support their network. The overall results of Jishnu Kinwar research were that uploading big files from Google Cloud Storage (AP) is 4x faster than AWS (3G). Also, uploading small chunks Google Cloud Storage (AP) is 20x faster than AWS S3 (SG), Google Cloud Storage costs 35% less than AWS for a similar solution. The upload mechanism used to upload to AWS was a Java based program which uses AWS API's whereas Google Cloud Storage uses Google's command line utility. And S3 media acceleration type feature by default. Consequently, that's a reason for faster uploads [10].

B. Accounts, limits, and pricing

User required to sign up for an AWS account to use AWS services. After login, AWS allow users to launch any service by following Amazon stated limits. Launched service charged to the user's account. Users can create billing accounts and then create sub-accounts that can appear. Correspondingly, GCS also required the user to set up an account to use their services but GCS charge for services by a project. Users can create various and exclusively distinctive projects on the same account. It's very helpful for testing as it allows users to delete the project once they complete the project, it will remove all resources created by that particular project. AWS and GCS have to default soft limits for new accounts on their services to prevent from risks from fraudulent accounts to use excessive resources. Prices of features and services tend to change more often. The price of these two Cloud Storage platforms can find one their websites [6]. AWS website is “https://calculator.s3.amazonaws.com/index.html”. GCS website is “https://cloud.google.com/products/calculator/."

C. Resource management interfaces

Each platform provides different CLI (Command-Line Interface) to interact with services and resources. Amazon Web Services provides "Amazon CLI, " and Google Cloud Storage offers "Cloud SDK." Both is cross-platform which implies that they're compatible with Windows, Mac OS X, and Linux. Cloud SDK can be used in a web browser using Google Cloud Shell. Both platforms also provide web-based consoles. This console enables users to create, manage, and monitor their resources [6].

D. Service types

The service types of both platforms include Distributed object storage, Reduced cost Data Storage, and Cold Data Storage [11]. Cloud allow storing a bunch of GB's to several PBs. The user interacts with APIs or third party programs. Google Cloud Storage provide a reliable and robust service, but resources are not even close to Amazon's [7].

The figure given below shows the service offered by both service providers and prices per month [7].

Figure 4: Service per GB/month

1. Distributed Object Storage

AWS S3 and GCS have hosted services to store and access to many binary items or blobs of variable sizes. Each service is highly scalable and redundant key-value store. In these services, keys are strings and values are arbitrary binary objects. These values and keys permit users to keep data objects. Each platform offers standard, reduced-cost, and cold tiers. Nonetheless, Amazon S3 and GCS offers distinct service classes which permit customers to exchange off amongst pace of retrieval and storage fee.

Each platform has similar service models. In each service, objects stored in a bucket where these objects are assigned a unique key. Each object has related metadata record which has information like media type, object size, and last modification date. Based on permissions, customers can view, upload, and regulate the metadata. Users are also allowed to feature custom designed metadata as demanded. AWS S3 and GCS persist the file system metaphor also by stipulating file system like APIs. Each service is also used to host static web content and media.

The figure shows a comparison of features and terminology of Amazon S3 and Google Cloud.

Figure 5: Distributed Object Storage

Indexing scheme used by Amazon S3 provide users with high throughput use cases obliged to be cautious approximately the manner they generate their object keys with the intention to prevent chances of potential hot spotting issues. Whereas, In Cloud storage, item name does now not influence the overall performance and scalability of gaining access to objects from the bucket.

Amazon S3 and GCS support item versioning. Each version stored with a given key stores with unique version identification number. While user permit versioning, it mitigates the danger of accidental data loss due to overwritten object.

In GCS, object versioning being for conditional updates like PUT and DELETE operations. If updated object version matched the designated object version within the request, then the update request will succeed only. Such method allows inhibiting the probability of race conditions all through updates. Whereas, Amazon S3 does not assist conditional updates.

Each platform permits users to automatize object deletion along with user selected lifecycle guidelines. In Amazon S3, during storage class users can set rules to migrate objects automatically.

While objects created, updated, or deleted, Amazon S3 and GCS issue notification. To issue warning users required to configure their buckets. This feature is referred to as by Amazon S3 "Event Notifications" while using Google Cloud Storage called as "Object Change Notifications." Amazon S3 comprise of Amazon SQS (Simple Queue Service) queue, Amazon SNS (Simple Notification Service), and AWS Lambda Function as destinations for notifications whereas Google Cloud Storage permit customers to post notifications to a targeted URL/Webhook. This targeted URL or Webhook handles a payload of notifications.

Each system provides with uptime guarantees and rules for credited customer's accounts if those guarantees are unobserved. These assurances and policies defined by both service providers in SLA (Service Level Agreement).

Amazon S3 and Google Cloud Storage charge by network egress and the amount of data stored per month. Furthermore, charge for common API requests [11].

The figure given below shows the summary of comparison of object storage of AWS S3 and Google Cloud Storage [12].

Figure 6: Summary of comparison of object storage of AWS S3 and Google Cloud Storage

2. Reduced Cost Data Storage

Reduced Cost Storage class for each platform do not require robustness of Standard Storage tier. Amazon S3 offers RRS (Reduced Redundancy Storage) which offer much less redundancy than standard tier while Google Cloud Storage gives DRA (Durable Reduced Availability) which provides less availability than standard tier.

Amazon S3 RRS and Cloud Storage DRA have a 1st-byte latency length of no more than a few milliseconds. The throughput of both storage class is same.

Both service providers provide with uptime guarantees and policies in the vicinity for credited client's debts if these assurances are not accompanied. Those guarantees and policies defined by both in SLA (Service Level Agreement) of Amazon S3 RRS and Google Cloud Storage DRA.

Amazon S3 and Google Cloud Storage charge by network egress and the amount of data stored per month. Moreover, it charges for standard API requests. Similarly, Amazon S3 RRS charge for data retrieval on per GB basis.

3. Cold Data Storage

Each Storage offers the same service for data that users do not require getting access to or retrieve on a regular basis. Amazon S3 provides Amazon S3 Standard (Infrequent Access Storage) whereas Google Cloud storage gives an additional class referred as Cloud Storage Nearline.

Amazon S3 and Google Cloud Storage have the least storage length for cold storage classes: Amazon S3 Standard (Infrequent Access Storage) data objects must remain unmodified for 30 days, and Cloud storage Nearline data objects should stay unmodified for 30 days. If a user deletes or overwrites data before minimal storage durations ends, a user might additionally charge for it.

The figure given below shows the minimum storage of each cold storage classes.

Figure 7: Each cold storage classes minimum storage.

Each platform provides uptime guarantees and policies in location for credited customer's accounts if these assurances are unobserved. Those guarantees and rules defined by both service providers in SLA (Service Level Agreement) of Amazon S3 preferred (Infrequent Access Storage) and Cloud storage Nearline.

Amazon S3 Standard (Infrequent Access Storage) charge by network egress and the amount of data stored per month. It has a 1st-byte latency period of several milliseconds. It has minimal object size of 128 KB objects smaller than 128 KB charged as if they may be 128 KB in length. If a user deletes or regulates data earlier than the minimum storage period, they will charge for the remainder of the durations. For example: If a user deletes an item after five days of storing the object then they may be charged for the remaining 25 days of storage for that particular object. Furthermore, costs for standard API requests and data retrieval on per GB basis.

Cloud Storage Nearline has a 1st-byte latency length of approx. Three seconds. Cloud Storage Nearline charge by network egress and the amount of data stored per month. If a user deletes or regulates data earlier than the minimum storage length, they cost for the rest of the intervals. For instance: If a user deletes an item after five days of storing the object then they'll be charged for the remaining 25 days of storage for that particular object. Furthermore, charge for common API requests.

E. Storage classes

All classes for AWS guarantee encryption at the server side. Also, allow lifecycle management and versioning while transferring [13].

Amazon S3 Storage has three storage classes:

a) Standard

• · Low latency

• · Easy access to stored objects

• · Recommended for urgent data

b) Standard-infrequent Access

• · Same services as Standard except it charge for data retrieval fee

c) Amazon Glacier

• Higher latency

• Lower Availability

• Data retrieval fee

• Recommended for prolonged data archiving.

Google Cloud Storage has three storage classes:

a) Standard

• High availability

• Low latency

b) Durable Reduced Availability

• Low availability

• Recommended to use for batch processes and similar states.

c) Nearline

• Low availability

• Higher latency

• Recommended to use for non-urgent storage of data

All classes for GCS enable to book redundant storage facilities and multi-regional preferences. They support data versioning on a bucket level. Server-side encryption performed for every situation. Stored files are compressed as needed. Every bucket permits to configure lifecycle settings for objects keep there. Furthermore, users of nearline storage class billed for early deletion or data retrieval [13].

The table given above shows the comparison between Amazon S3 and Google Cloud Storage classes features [13].

Figure 9: Storage classes

A. Cloud Storage Performance

Google Cloud Storage and Amazon S3 allow worldwide storage and retrieval of any amount of data at any time [5] [3]. Object storage of Amazon AWS S3 and Google Cloud Storage offers a simple PUT/GET/HEAD/LIST interface to store data in the database. It is cheap and provides safety via scales and redundancy factor automatically to serve various simultaneous requests. The drawbacks are throughput and latency due to being restricted by network availability and performance.

To download files from a Virtual Machine hosted in the same region by the same vendor then two key metrics can be considered: TTFB (Time to First Byte) to evaluate latency, and throughput. According to research, Google Cloud Storage averaged more than three times higher latency as compare to AWS S3. Conversely, Google provided approx. 2x the throughput of S3. Therefore, Google Cloud Storage can finish downloading quicker than S3 for files larger than 5 MB.

Multi-regional storage data is geo-redundant which don't allow the user to store regional storage data in a bucket created in a multi-regional location [14]. S3 buckets and GCS regional buckets are equivalent. Multi-regional buckets enable access from a set of regions without any data transfer charges or duplication costs whereas regional bucket provides lower latency and better output than a multi-regional bucket. Google also have higher latency, but faster throughput and AWS are quite less. Object storage handles concurrent requests well as Googles “gsuntil” comprise of “Perfdiag” command can be used by GCS and S3 as its built on boto. This tool allows to read and write multiple objects at the same time. Hence, it can serve concurrent requests [15].

Amazon S3 us faster storage depends on user's requirements. Use cases for data archive are backup tape replacement, enterprise compliance, and media asset archival. Backup tape replacement makes the backup process manageable and provides better cost savings and efficiency. Enterprise compliance includes an assurance that user's data stored in AWS meets all the requirements for their business. Media asset archival allow users to store digital assets in encrypted and redundant storage. Amazon S3 support secure transfer of data over SSL (Secure Sockets Layer). It can automatically encrypt the data using AES (Advanced Encryption Standard) 256-bit symmetric keys. AWS simplify the data transfer as it enables data transfer in a convenient way. It provides a cost saving to move data from S3. The AWS IAM (Identity and Access Management) service allow users to control and manage access to data. It gives them the power to authenticate users to what resources and service they can access. Direct Connect offers the user with dedicated private network connections for private, secure, and dedicated access to AWS cloud and facilities. It delivers Increase durability of 99.999999999% through synchronously storing user's data within various services [16].

AWS offer reliable, durable, and highly-scalable cloud storage for backup and are designed to assist mission- critical databases like SAP and Oracle [17]. AWS uses classical backup facilities which offer snapshot facility where incremental backups are taken from the data store every few minutes which result in most inclusive and up to date restore environment [18].

GCS backup offers a way to restore Cloud SQL Instance recover lost or corrupt data. Users required to enable automated backup for any instance that comprise of data what they want to protect from any loss or damage. Some operations required enabling automatic backup amidst binary logging. For instance: replica creation and clone. Each instance cloud SQL preserves more than seven automated backups [19]. Google Cloud SQL prices depend on instance types: MySQL second generation, PostgreSQL, and MySQL first generation [20]. GCS uses incremental backup which comprises of data changed. The size of oldest backup will be like user's database, but the size of subsequent backup will depend on the rate of change of user's data. If the user deletes the old backup, then the size of next backup will increase, so that full back will still exist. The user needs to enable the automatic backup which allows 4 hour's backup window then it will begin to backup during the backup window. When users instance has the least activity, it will possibly schedule backup. If there are no changes, it means there is no backup taken. GCS also provides about on-demand backups. It allows users to create a backup at any time for second generation instances if they are performing a risky operation on their database. Hence, they don't need to wait for the backup window. It doesn't require enabling the automatic backup. It doesn't delete automatically like automatic database as they persist until the user deletes them. It will affect user's storage charges. CloudSQL allows users to create an individual user "cloudsqladmin" for each instance and generates the unique instance-specific password for it. Automatic backup performed by login as "cloudsqladmin."

The backups affect instance operations. For first generation instance, backups acquired by using flag "FLUSH TABLES WITH READ LOCK" to create a snapshot. It blocks the instance during backup. The instance remains online, and reads will be unaffected. Backup takes few seconds to complete, but if a significant amount of data written as the previous backup, then the backup will take longer to complete. A pending operations at the time of the backup attempt, CloudSQL makes numerous attempts within the window to finish the backup. Long-running Operations like import, export, update, restart (instance) blocks the backup or temporary disable the automated backups. For second generation instances, it doesn't use the flag as used in first generation instances. Hence, writes and other operations are unaffected by backup operations [19].

B. Migration

1) Migration from Amazon S3 to Google Cloud Storage

Amazon S3 allow users to migrate their applications to use Google Cloud Storage efficiently. There are two migration options:

Simple migration: Few changes made in tools and libraries user is currently using with Amazon S3. All features of Google Cloud Storage cannot be used using simple migration.

Full migration: More changes are required as compare to simple migration. All features of Google Cloud Storage can be utilized incorporating multiple projects, service accounts, and OAuth 2.0 for authentication [21].

2) Migrating from Google Cloud Storage to Amazon S3

Data can be migrating from Google Cloud Storage to Amazon using following options:

Create virtual server using Google Compute Engine account. Data extracted from Google Cloud Storage into compute engine server and migrate to S3 bucket. User required configured permission from GCS and AWS S3 bucket.

An EC2 instance created on AWS, and same activities performed over there. The user needed set permission from GCS and AWS S3 bucket.

Download and install third-party tools on the local machine which can be used to perform the operation of downloading GCS data into the local machine and uploading to AWS S3 bucket. However, this option will be slower as it requires downloading first on local machine [22].

Minio Client “mc” can be used to migrate data as it supports both GCS and AWS S3. Its features include upload, downloads, progress bar [23].

Hadoop Distcp also can use to move a large amount of data. 3-node Hadoop cluster can be created using Google Dataproc [24].

Conclusion

As I Concluded, it's difficult for users to select a platform which offers more services. To select best option comparison are performed to provide users to choose the best choice. Amazon Web Service provide few option while user signup for their services whereas Google Cloud platform provide with $300 to use any of its services using that credit. Amazon S3 services are implemented within region location globally whereas Google Cloud Storage available in multi-regional locations. According to research conducted by Jishnu Kinwar, uploading big files from Google Cloud Storage (AP) is 4x faster than AWS (3G). Also, uploading small chunks Google Cloud Storage (AP) is 20x faster than AWS S3 (SG), Google Cloud Storage costs 35% less than AWS for a similar solution. In AWS users are charged based on launched service whereas Google users are charged based on per project. Each platform provides different CLI (Command-Line Interface) to interact with services and resources. Google Cloud Storage provide a reliable and robust service, but resources are not even close to Amazon's. The distributed object storage feature for both are similar just few difference noticed like Update notifications, Service classes, and Deployment locality. Amazon S3 update notifications are called as "Event notifications" whereas Google Cloud Storage called as "Object change notifications." Service classes for Amazon S3 are Standard, Reduced Redundancy, Infrequent Access, and Amazon Glacier whereas for Google Cloud Storage is Standard, Durable Reduced Availability, Nearline. According to research, Google Cloud Storage averaged more than three times higher latency as compare to AWS S3. It observed that Amazon S3 allow users to migrate their applications to use Google Cloud Storage efficiently whereas for Google Cloud Storage there is not such option, but it can be migrated using methods given in the migration section. Hence, Google Cloud Storage is more recommendable than Amazon S3.

References

[1] C. Cybrary, University of North Carolina at Chapel Hill Libraries, 7 2015.[Online].Available:http://library.unc.edu/wp-content/uploads/2014/04/Cloud_Storage.pdf. [Accessed 29 3 2017].

[2] API, "Amazon Simple Storage Service," Amazon Simple Storage Service, 2006.

[3] Cloud.google,2017.[Online].Available: https://cloud.google.com/storage/docs/. [Accessed 4 4 2017].

[4] S. B. Margaret Rouse, "Google Cloud Platform," pp. 3-8, 1 2016.

[5] AWS, 2017. [Online]. Available: https://aws.amazon.com/s3/. [Accessed 4 4 2017].

[6] G. C. P. f. A. Professionals, "cloud.google," cloud.google, 29 6 2016. [Online]. Available: https://cloud.google.com/docs/compare/aws/. [Accessed 7 4 2017].

[7] L. d. Alba, A Side-by-Side Comparison of AWS, Google Cloud and Azure, 2016, pp. 3-8.

[8] A. G. Infrastructure, "aws," 2017. [Online]. Available: https://aws.amazon.com/about-aws/global-infrastructure/. [Accessed 4 4 2017].

[9] C. Locations, "GCS," GCS, 2017. [Online]. Available: https://cloud.google.com/about/locations/. [Accessed 4 4 2017].

[10] J. Kinwar, Google Cloud Storage Vs AWS S3, LinkedIn, 2016.

[11] Storage, "Cloud.google," 29 7 2016. [Online]. Available: https://cloud.google.com/docs/compare/aws/storage#distributed_object_storage. [Accessed 9 4 2017].

[12] B. Adler, rightscale, 18 4 2016. [Online]. Available: http://www.rightscale.com/blog/cloud-management-best-practices/cloud-storage-aws-vs-azure-vs-google-vs-softlayer. [Accessed 4 4 2017].

[13] V. S, Compare AWS, Microsoft Azure and Google Cloud for Backup, Amazon, 2016, pp. 1-3.

[14] GCS, Multi-regiona, cloud storage, pp. 1-6.

[15] Z. Bjornson, AWS S3 vs. Google Cloud vs. Azure:, Zach Bjornson, 2015.

[16] D. Archive, Amazon, 2017. [Online]. Available: https://aws.amazon.com/archive/. [Accessed 4 4 2017].

[17] Backup-Recovery, 2017. [Online]. Available: https://aws.amazon.com/backup-recovery/. [Accessed 4 4 2017].

[18] I. solutions, 2011. [Online]. Available: http://www.1t-s.com/data-protection-backup-restore. [Accessed 4 4 2017].

[19] Backups, GCS, 2017. [Online]. Available: https://cloud.google.com/sql/docs/mysql/backup-recovery/backups. [Accessed 4 4 2017].

[20] Pricing, Amazon, 2017. [Online]. Available: https://cloud.google.com/sql/pricing#2nd-gen-storage-networking-prices. [Accessed 4 4 2017].

[21] Migrating, "GCS," 2017. [Online]. Available: https://cloud.google.com/storage/docs/migrating. [Accessed 6 4 2017].

[22] H. Sachdeva, How can I migrate data from Google cloud storage into AWS S3 buckets?, Bengaluru, Karnataka, India: Quora, 201.

[23] Harshavardhana, How can I migrate data from Google cloud storage into AWS S3 buckets?, Palo Alto, CA: Quora, 2016.

[24] S. Hermans, How can I migrate data from Google cloud storage into AWS S3 buckets?, Papillion, NE: Quora, 2017.