Bigstep vs GCP

Bigstep Metal Cloud vs GCP

Among the top 3 competitors we like Google's services the most. However we have our advantages which make the Metal Cloud a solid alternative. We've put together a side-by-side comparison to make it easy to make a decision.

	Bigstep Metal Cloud	GCP
Bare metal	Yes	No
Minimum instance size	8 cores 32 GB RAM	1 core 3.75GB RAM
Self Service	Yes	Yes
Provisioning time	300 seconds	100-300 seconds
OS volume	iscsi	via "persistent disks"
Additional storage volumes	iscsi block devices	via "persistent disks"
On-demand	Yes (per second)	Yes (per second)
Migration-less upgrade support	Yes (with restart)	Yes (including live migration)
Stop and resume support	Yes	Yes
Service network config.	Yes (L2 & L3)	Yes (L3 only)
L2 technology	MPLS	N/A
Private LAN support	Yes	Yes (VPC)
API enabled	Yes	Yes
IPV6	Yes	Yes (only as HTTP(S),SSl termination)
BGP support	Yes (custom)	Yes (native)
Shared filesystem support	Yes (via GFS -Bigstep SharedDrive)	Yes (via Cloud Filestore)
Database as a Service	Couchbase, Cassandra, Exasol, PostgreSQL, MySQL (Percona), Elasticsearch	CloudSQL, Cloud BigTable, Cloud Spanner, Cloud Firestore, Cloud Datastore, Cloud Memorystore
Hadoop & Spark as a Service	Cloudera, MapR, Hortonworks, Mesos	Dataproc
Data Science Tools	Jupyter	Colaboratory, DataLab
Native container platform	Yes (Bigstep Container Platform)	Yes (GKE)
Container Apps	Apache Spark, Streamsets, Apache Kafka, Zoomdata, Jupyter	No
Object Storage	Yes (Bigstep DataLake service)	Yes (Cloud Storage)
Serverless services	No	Cloud Functions, AppEngine, DataFlow, Pub/Sub, BigQuery
Other services	No	Many

Why choose Bigstep?

Single tenant, bare metal
at no extra cost

Our primary advantage comes from the performance and stability improvements that bare metal offers. You can get "sole-tenant" servers in GCP but at a 10% premium and you will still have the virtualization layer.

2x compute performance

The Metal Cloud's typical performance improvement you should expect is 2x although it depends on the kind of workload you have. A customer found 20x improvement for deep learning model training.

6x lower network latency

The network latency is 6x lower in Bigstep than in GCP so microservices on Kubernetes will run faster in Bigstep than GCP.

Why choose GCP?

Serverless &
shared hosting services

We highlight their AppEngine which is an modern shared hosting environment designed to run apps written in many languages. That is ideal for hobby or otherwise small projects and we recommend them to developers that are just getting started.

Just like AWS, GCP has a service for everything, ranging from publish-subscribe queues and other. Most of them are wrappers around existing technologies such as Kafka but some are Google own, unique IP that is not open source (eg: BigQuery).

Closing words on managed services

The shier broadness of managed services might incline the user to opt for the "just in case I need it" scenario but one must remember that through open source (or commercial) alternatives most of the services can be easily replicated and managed by a 3rd party if the ops resources are not available. What that offers is (sometimes much) lower cost, vendor independence and control.

As cloud services become comoditized, portability will be essential in being able to quickly leverage future cost reductions or stability issues.