Multi-Tenancy Architecture

Leave a Comment / Blog / By

What Is Multi-Tenancy?

Mul­ti-ten­an­cy is a fun­da­men­tal archi­tec­ture in cloud com­put­ing where a sin­gle instance of a soft­ware appli­ca­tion serves mul­ti­ple cus­tomers, known as ten­ants. This archi­tec­ture allows for shar­ing of resources, such as data­bas­es and servers, among dif­fer­ent users while keep­ing their data sep­a­rate and secure. Mul­ti-ten­an­cy is a key char­ac­ter­is­tic of many cloud ser­vices, par­tic­u­lar­ly in Soft­ware as a Ser­vice (SaaS) mod­els, opti­miz­ing resource uti­liza­tion and reduc­ing costs.

What Does Multi-Tenancy Include?

Mul­ti-ten­an­cy includes sev­er­al key com­po­nents and char­ac­ter­is­tics, such as shared resources, data iso­la­tion, scal­a­bil­i­ty, cen­tral­ized man­age­ment, cost-effec­tive­ness, and cus­tomiza­tion. All of these com­po­nents work togeth­er to help achieve the desired out­come.

  1. Shared Resources: The core of mul­ti-ten­an­cy is the shar­ing of resources. This includes infra­struc­ture, appli­ca­tions, and data­bas­es which are shared by mul­ti­ple ten­ants.
  2. Data Iso­la­tion: Despite the shared envi­ron­ment, each ten­an­t’s data is iso­lat­ed and remains invis­i­ble to oth­er ten­ants. This is achieved through data­base schema or mul­ti-instance archi­tec­ture.
  3. Scal­a­bil­i­ty: Mul­ti-ten­ant archi­tec­tures are inher­ent­ly scal­able, as they allow for adding new ten­ants with­out sig­nif­i­cant changes to the infra­struc­ture.
  4. Cen­tral­ized Man­age­ment: The main­te­nance and updat­ing of these sys­tems are cen­tral­ized, mean­ing that soft­ware updates or patch­es need to be done only once and are then avail­able to all ten­ants.
  5. Cost-Effec­tive­ness: By pool­ing resources, mul­ti-ten­an­cy can reduce costs, both for the provider and the ten­ants, by max­i­miz­ing the effi­cien­cy of resource uti­liza­tion.
  6. Cus­tomiza­tion: Many mul­ti-ten­ant archi­tec­tures allow for a degree of cus­tomiza­tion, enabling ten­ants to per­son­al­ize aspects of the appli­ca­tion to suit their spe­cif­ic needs.

Who Primarily Uses Multi-Tenancy?

Mul­ti-ten­an­cy is pri­mar­i­ly used by SaaS providers, busi­ness­es of all sizes, edu­ca­tion insti­tu­tions and non-prof­it orga­ni­za­tions, and gov­ern­ment agen­cies. While each of these orga­ni­za­tions is vast­ly dif­fer­ent, they all ben­e­fit from the mul­ti-ten­an­cy archi­tec­ture.

  1. SaaS Providers: They use mul­ti-ten­an­cy to effi­cient­ly offer their soft­ware ser­vices to a wide range of cus­tomers. Exam­ples include CRM sys­tems, busi­ness man­age­ment tools, and email ser­vices.
  2. Busi­ness­es of All Sizes: From small star­tups to large enter­pris­es, busi­ness­es lever­age mul­ti-ten­ant SaaS solu­tions for var­i­ous func­tions due to their cost-effec­tive­ness and scal­a­bil­i­ty.
  3. Edu­ca­tion­al Insti­tu­tions and Non-Prof­its: These orga­ni­za­tions often uti­lize mul­ti-ten­ant plat­forms for col­lab­o­ra­tion tools, doc­u­ment man­age­ment, and oth­er appli­ca­tion ser­vices that require scal­a­bil­i­ty and are cost-sen­si­tive.
  4. Gov­ern­ment Agen­cies: These agen­cies use mul­ti-ten­ant cloud ser­vices for var­i­ous pub­lic ser­vices, ben­e­fit­ing from the cost sav­ings and effi­cien­cy it offers.

Types of Multi-Tenant Architecture

Mul­ti-ten­ant archi­tec­ture is a cor­ner­stone in cloud com­put­ing, enabling effi­cient resource shar­ing while main­tain­ing ten­ant iso­la­tion and secu­ri­ty. There are sev­er­al types of mul­ti-ten­ant archi­tec­tures, each with its unique approach to resource shar­ing and data man­age­ment. Under­stand­ing these vari­a­tions is cru­cial for busi­ness­es and soft­ware devel­op­ers to choose the most appro­pri­ate mod­el for their spe­cif­ic needs.

1. Shared Data­base, Shared Schema

In the “shared data­base, shared schema” mod­el, all ten­ants share both the same data­base and the same data­base schema. This set­up is the most inte­grat­ed form of mul­ti-ten­an­cy and is char­ac­ter­ized by the fol­low­ing:

  • Resource Opti­miza­tion: This mod­el achieves the high­est lev­el of resource shar­ing, lead­ing to opti­mized use of serv­er and stor­age resources.
  • Uni­for­mi­ty: All ten­ants share the same data­base struc­ture, which sim­pli­fies main­te­nance and updates.
  • Data Iso­la­tion: It uses ten­ant-spe­cif­ic iden­ti­fiers with­in the data­base to dis­tin­guish and secure­ly sep­a­rate data for each ten­ant.
  • Cost-Effec­tive­ness: Due to its high degree of shar­ing, it’s often the most cost-effec­tive mod­el for providers.

How­ev­er, this approach can have lim­i­ta­tions in terms of cus­tomiza­tion and scal­a­bil­i­ty for indi­vid­ual ten­ants, as all ten­ants are tight­ly inte­grat­ed into the same data­base struc­ture.

2. Shared Data­base, Mul­ti­ple Schemas

The “shared data­base, mul­ti­ple schemas” mod­el involves a sin­gle data­base instance, but each ten­ant has its own data­base schema.

  • Mod­er­ate Iso­la­tion: This approach offers a bet­ter iso­la­tion lev­el than the shared schema mod­el, as each ten­an­t’s data and data­base struc­ture are kept sep­a­rate.
  • Cus­tomiza­tion: It allows for some lev­el of cus­tomiza­tion in the data­base schema for each ten­ant.
  • Effi­cien­cy in Resource Use: While it pro­vides more iso­la­tion, it still main­tains effi­cien­cy in terms of resource use and cost.
  • Main­te­nance Bal­ance: Strikes a bal­ance between resource shar­ing and the abil­i­ty to cater to dif­fer­ent ten­ant needs.

This mod­el is gen­er­al­ly suit­able for sce­nar­ios where ten­ants require some lev­el of cus­tomiza­tion while still ben­e­fit­ing from the effi­cien­cies of shared infra­struc­ture.

3. Mul­ti­ple Data­bas­es, Mul­ti­ple Schemas

In the “mul­ti­ple data­bas­es, mul­ti­ple schemas” mod­el, each ten­ant has its own data­base instance and schema.

  • Max­i­mum Iso­la­tion: This offers the high­est lev­el of data iso­la­tion and secu­ri­ty among the mul­ti-ten­ant archi­tec­ture types.
  • Cus­tomiza­tion and Flex­i­bil­i­ty: This enables sig­nif­i­cant cus­tomiza­tion capa­bil­i­ties for each ten­ant, both in terms of data man­age­ment and schema design.
  • Resource Allo­ca­tion: Each ten­an­t’s data­base can be scaled inde­pen­dent­ly, offer­ing greater con­trol over resource allo­ca­tion.
  • High­er Costs: This mod­el tends to be more resource-inten­sive and can be more expen­sive due to the need for more infra­struc­ture and main­te­nance efforts.

This approach is often pre­ferred in sit­u­a­tions where ten­ants require exten­sive cus­tomiza­tion, have large-scale oper­a­tions, or when data secu­ri­ty and iso­la­tion are top pri­or­i­ties.

Each type of mul­ti-ten­ant archi­tec­ture offers dif­fer­ent lev­els of data iso­la­tion, cus­tomiza­tion, and cost effi­cien­cy. The choice of archi­tec­ture depends on the spe­cif­ic require­ments of the ser­vice provider and their ten­ants, bal­anc­ing between resource opti­miza­tion, ten­ant iso­la­tion, and the abil­i­ty to cus­tomize.

Pros of Multi-Tenancy

Mul­ti-ten­an­cy in cloud com­put­ing offers a range of advan­tages that can sig­nif­i­cant­ly ben­e­fit busi­ness­es in terms of cost sav­ings, data pro­cess­ing effi­cien­cies, and scal­a­bil­i­ty. Under­stand­ing these ben­e­fits is cru­cial for orga­ni­za­tions con­sid­er­ing adopt­ing a mul­ti-ten­ant archi­tec­ture for their cloud-based ser­vices.

1. Helps Save Mon­ey

One of the pri­ma­ry advan­tages of mul­ti-ten­an­cy is its poten­tial for cost sav­ings. Rang­ing from reduced infra­struc­ture costs to effi­cient use of resources, mul­ti-ten­an­cy pro­vides mul­ti­ple areas in which orga­ni­za­tions can cut costs with­out sac­ri­fic­ing qual­i­ty.

  • Reduced Infra­struc­ture Costs: By shar­ing under­ly­ing infra­struc­ture among mul­ti­ple ten­ants, busi­ness­es can sig­nif­i­cant­ly reduce the costs asso­ci­at­ed with pur­chas­ing and main­tain­ing hard­ware and soft­ware resources.
  • Economies of Scale: Mul­ti-ten­ant envi­ron­ments ben­e­fit from economies of scale, where the cost per ten­ant decreas­es as more ten­ants use the sys­tem.
  • Low­er Oper­a­tional Costs: Shared resources means that oper­a­tional tasks such as main­te­nance, updates, and back­ups can be cen­tral­ized, fur­ther reduc­ing the cost and effort required for these activ­i­ties.
  • Effi­cient Resource Uti­liza­tion: Mul­ti-ten­an­cy opti­mizes the use of avail­able resources, ensur­ing that they are not under­uti­lized, which con­tributes to cost effi­cien­cy.

2. Helps Process Data

Because mul­ti-ten­an­cy pro­vides var­i­ous ways to man­age data, it helps to keep data cen­tral­ized, improve data insights, and stream­line updates. Pro­cess­ing data is enhanced because of the fol­low­ing capa­bil­i­ties:

  • Cen­tral­ized Man­age­ment: Data from all ten­ants can be man­aged cen­tral­ly, sim­pli­fy­ing tasks such as data analy­sis, back­ups, and recov­ery.
  • Improved Data Insights: With all ten­ant data stored in a uni­fied sys­tem, it becomes eas­i­er to per­form com­pre­hen­sive ana­lyt­ics, offer­ing bet­ter insights and help­ing in deci­sion-mak­ing process­es.
  • Stream­lined Updates: Any improve­ments or updates to the data pro­cess­ing mech­a­nisms can be imple­ment­ed across all ten­ants simul­ta­ne­ous­ly, ensur­ing that all ben­e­fit from the lat­est func­tion­al­i­ties.

3. Helps Scal­a­bil­i­ty 

Being able to scale and grow is essen­tial to any orga­ni­za­tion, and mul­ti-ten­an­cy pro­vides ways to help with scal­a­bil­i­ty. Scal­a­bil­i­ty is evi­dent in the fol­low­ing areas of a mul­ti-ten­an­cy archi­tec­ture:

  • Flex­i­bil­i­ty in Resource Allo­ca­tion: Mul­ti-ten­an­cy allows for dynam­ic allo­ca­tion of resources based on demand, mak­ing it eas­i­er to scale up or down as required.
  • Han­dling Growth: As the num­ber of ten­ants increas­es, the sys­tem can be scaled to accom­mo­date this growth with­out sig­nif­i­cant changes to the infra­struc­ture.
  • Adapt­abil­i­ty to Chang­ing Needs: Mul­ti-ten­ant sys­tems are designed to be adapt­able, allow­ing busi­ness­es to respond quick­ly to changes in demand or usage pat­terns.
  • Cost-Effec­tive Scal­ing: Scal­ing resources in a mul­ti-ten­ant envi­ron­ment is gen­er­al­ly more cost-effec­tive than in sin­gle-ten­ant sys­tems, as the infra­struc­ture costs are shared among all ten­ants.

Cons of Multi-Tenancy

While mul­ti-ten­an­cy offers sev­er­al advan­tages, it also comes with its own set of chal­lenges and lim­i­ta­tions. Under­stand­ing these draw­backs is essen­tial for orga­ni­za­tions to make informed deci­sions and man­age poten­tial issues effec­tive­ly when adopt­ing a mul­ti-ten­ant archi­tec­ture.

1. Com­plex Devel­op­ment

While a mul­ti-ten­ant archi­tec­ture does pro­vide room for scal­a­bil­i­ty, the com­plex­i­ty of devel­op­ing a mul­ti-ten­ant archi­tec­ture is one of its sig­nif­i­cant dis­ad­van­tages. There are some tech­ni­cal aspects of this archi­tec­ture that are required to achieve a desired out­come:

  • Intri­cate Archi­tec­ture Design: Design­ing a mul­ti-ten­ant archi­tec­ture requires care­ful plan­ning to ensure data iso­la­tion, secu­ri­ty, and effi­cient resource shar­ing, which can be com­plex and time-con­sum­ing.
  • Advanced Tech­ni­cal Skills: Devel­op­ers need a high lev­el of exper­tise in cloud com­put­ing and data­base man­age­ment to imple­ment a robust mul­ti-ten­ant sys­tem.
  • Strin­gent Secu­ri­ty Require­ments: Ensur­ing data secu­ri­ty in a shared envi­ron­ment demands sophis­ti­cat­ed secu­ri­ty mea­sures, adding to the com­plex­i­ty.
  • Cus­tomiza­tion Chal­lenges: Design­ing a sys­tem that meets the diverse needs of mul­ti­ple ten­ants while main­tain­ing a stan­dard struc­ture can be chal­leng­ing.

2. Lim­it­ed Per­son­al­iza­tion

There is great ben­e­fit in being able to use resources effi­cient­ly; how­ev­er, it does come at a price. Because of some of the lim­i­ta­tions of this archi­tec­ture, mul­ti-ten­an­cy can restrict the lev­el of per­son­al­iza­tion that is achiev­able for each ten­ant:

  • Stan­dard­ized Fea­tures: In a mul­ti-ten­ant envi­ron­ment, the application’s core fea­tures are stan­dard­ized across all ten­ants, which may not meet spe­cif­ic indi­vid­ual require­ments.
  • Lim­it­ed Cus­tomiza­tion Options: Ten­ants often have lim­it­ed abil­i­ty to cus­tomize the soft­ware to their unique busi­ness process­es or brand­ing require­ments.
  • One-Size-Fits-All Approach: This approach can some­times lead to a com­pro­mise on the part of ten­ants who might need more tai­lored solu­tions.

3. Com­pli­cat­ed Changes

Hav­ing cen­tral­ized data man­age­ment does allow for easy access to all ten­ants, but mak­ing changes to the struc­ture does not allow for an easy fix. Imple­ment­ing changes in a mul­ti-ten­ant archi­tec­ture can be com­pli­cat­ed due to its shared nature:

  • Impact on All Ten­ants: Any changes or updates affect all ten­ants, which requires care­ful plan­ning and test­ing to avoid dis­rup­tions or unin­tend­ed con­se­quences.
  • Coor­di­nat­ing Updates: Sched­ul­ing and imple­ment­ing updates across mul­ti­ple ten­ants can be logis­ti­cal­ly chal­leng­ing, espe­cial­ly when down­time needs to be min­i­mized.
  • Bal­anc­ing Diverse Needs: Accom­mo­dat­ing the vary­ing needs and pref­er­ences of dif­fer­ent ten­ants when mak­ing changes can be dif­fi­cult, as what ben­e­fits one ten­ant might not suit anoth­er.
  • Regres­sion Test­ing: Exten­sive test­ing is required to ensure that changes do not adverse­ly affect the func­tion­al­i­ty or per­for­mance for any ten­ant.

In sum­ma­ry, while mul­ti-ten­an­cy is a cost-effec­tive and effi­cient approach for cloud-based solu­tions, it does come with chal­lenges relat­ed to devel­op­ment com­plex­i­ty, lim­it­ed per­son­al­iza­tion, and the intri­ca­cies involved in imple­ment­ing changes. Orga­ni­za­tions should weigh these cons against the pros to deter­mine if a mul­ti-ten­ant archi­tec­ture aligns with their oper­a­tional needs and capa­bil­i­ties.

Multi-Tenant Architecture Examples

Mul­ti-ten­ant archi­tec­ture, a key con­cept in cloud com­put­ing and SaaS, comes in var­i­ous forms, each designed to cater to spe­cif­ic needs and sce­nar­ios. Under­stand­ing these exam­ples is essen­tial for busi­ness­es and soft­ware devel­op­ers to choose the right mod­el for their ser­vices.

1. URL-Based SaaS

URL-based SaaS is a form of mul­ti-ten­ant archi­tec­ture where each ten­ant access­es the ser­vice through a unique URL. This mod­el is par­tic­u­lar­ly com­mon in web-based appli­ca­tions and ser­vices.

  • Dis­tinct Access Points: Ten­ants access their instance of the appli­ca­tion via a unique web address, which may be cus­tomized to their busi­ness or brand.
  • Shared Infra­struc­ture: Despite the unique URLs, all ten­ants share the same under­ly­ing infra­struc­ture, which opti­mizes resource use and low­ers costs.
  • Data Seg­re­ga­tion: Although shar­ing the same appli­ca­tion, data for each ten­ant is seg­re­gat­ed and secure, ensur­ing pri­va­cy and secu­ri­ty com­pli­ance.
  • Ease of Use: It’s user-friend­ly, as ten­ants can eas­i­ly remem­ber and access their spe­cif­ic URL, enhanc­ing the user expe­ri­ence.

2. Mul­ti-Ten­ant SaaS

Mul­ti-ten­ant SaaS is a broad cat­e­go­ry encom­pass­ing var­i­ous SaaS appli­ca­tions that serve mul­ti­ple ten­ants from a sin­gle instance of the soft­ware.

  • Resource Effi­cien­cy: This mod­el max­i­mizes the effi­cien­cy of resource usage by serv­ing mul­ti­ple cus­tomers with the same appli­ca­tion set­up.
  • Cost-Effec­tive: By dis­trib­ut­ing the cost of main­te­nance and infra­struc­ture across mul­ti­ple ten­ants, it reduces over­all expens­es for both the provider and the ten­ants.
  • Uni­form Fea­tures:  Because of a stan­dard­ized set of fea­tures to all ten­ants, the devel­op­ment and update process­es are stream­lined. 
  • Scal­a­bil­i­ty: This mod­el is eas­i­ly scal­able, as adding new ten­ants doesn’t require addi­tion­al instances of the soft­ware.

3. Vir­tu­al­iza­tion-Based SaaS

Vir­tu­al­iza­tion-based SaaS uses vir­tu­al­iza­tion tech­nol­o­gy to cre­ate sep­a­rate, iso­lat­ed envi­ron­ments for each ten­ant with­in the same serv­er.

  • Enhanced Iso­la­tion: This mod­el pro­vides a high­er degree of iso­la­tion com­pared to oth­er mul­ti-ten­ant mod­els, as each ten­ant oper­ates in a sep­a­rate vir­tu­al envi­ron­ment.
  • Cus­tomiza­tion Flex­i­bil­i­ty: It offers more flex­i­bil­i­ty for cus­tomiza­tion and con­fig­u­ra­tion to meet spe­cif­ic ten­ant needs.
  • Resource Allo­ca­tion: Because it allows for more pre­cise con­trol over resource allo­ca­tion and man­age­ment, each ten­ant can opti­mize their per­for­mance.
  • Secu­ri­ty and Pri­va­cy: Secu­ri­ty and pri­va­cy are improved, as the vir­tu­al­ized envi­ron­ments are dis­tinct and iso­lat­ed from each oth­er.

Each of these mul­ti-ten­ant archi­tec­ture exam­ples has its unique fea­tures and ben­e­fits, mak­ing them suit­able for dif­fer­ent kinds of SaaS appli­ca­tions and cus­tomer require­ments. The choice depends on fac­tors such as the lev­el of cus­tomiza­tion need­ed, secu­ri­ty require­ments, and the spe­cif­ic oper­a­tional needs of the ten­ants.

What Is the Difference Between a Multi-Tenant and Single-Tenant Architecture?

In the realm of cloud com­put­ing and SaaS, the dis­tinc­tion between mul­ti-ten­ant and sin­gle-ten­ant archi­tec­tures is cru­cial. Each has its unique attrib­ut­es and impli­ca­tions for costs„ effi­cien­cy, and resource access. Under­stand­ing these dif­fer­ences helps busi­ness­es make informed deci­sions about which archi­tec­ture best suits their needs.

1. Cost Com­par­i­son

  • Mul­ti-Ten­ant Archi­tec­ture: Gen­er­al­ly more cost-effec­tive due to shared resources among mul­ti­ple clients. The economies of scale allow for low­er oper­a­tional costs, and sav­ings are often passed on to the clients. Main­te­nance and update costs are also shared, fur­ther reduc­ing expens­es.
  • Sin­gle-Ten­ant Archi­tec­ture: Typ­i­cal­ly incurs high­er costs, as it requires ded­i­cat­ed resources for each client. This includes the costs of main­tain­ing sep­a­rate instances of the soft­ware and the under­ly­ing hard­ware. While offer­ing exclu­sive use, it demands a high­er invest­ment from the client.

2. Effi­cien­cy Com­par­i­son

  • Mul­ti-Ten­ant Archi­tec­ture: Tends to be more effi­cient in terms of resource uti­liza­tion. Shared resources mean that the infra­struc­ture can be opti­mized accord­ing to col­lec­tive demand, min­i­miz­ing waste.
  • Sin­gle-Ten­ant Archi­tec­ture: Effi­cien­cy can vary. While resources are exclu­sive­ly avail­able to one client, this can lead to under­uti­liza­tion, espe­cial­ly if the clien­t’s demand fluc­tu­ates.

 3. Hard­ware Resource Access Com­par­i­son

  • Mul­ti-Ten­ant Archi­tec­ture: Hard­ware resources are shared among mul­ti­ple ten­ants, which means that access can be lim­it­ed based on over­all demand and usage poli­cies. How­ev­er, mod­ern cloud ser­vices are adept at man­ag­ing these resources to pro­vide suf­fi­cient access for all ten­ants.
  • Sin­gle-Ten­ant Archi­tec­ture: Clients have exclu­sive access to hard­ware resources, which can be advan­ta­geous for per­for­mance-inten­sive appli­ca­tions or when spe­cif­ic hard­ware con­fig­u­ra­tions are required.

4. Soft­ware Resource Access Com­par­i­son

  • Mul­ti-Ten­ant Archi­tec­ture: All ten­ants access the same soft­ware instance, which ensures uni­for­mi­ty in fea­tures and func­tion­al­i­ty. Cus­tomiza­tion options might be lim­it­ed com­pared to sin­gle-ten­ant solu­tions.
  • Sin­gle-Ten­ant Archi­tec­ture: Offers the oppor­tu­ni­ty for more exten­sive cus­tomiza­tion and con­trol over the soft­ware envi­ron­ment. This can be ben­e­fi­cial for clients with unique or spe­cial­ized soft­ware require­ments.

When choos­ing the archi­tec­ture that works best, con­sid­er the options that fit your orga­ni­za­tion bet­ter. Whether mul­ti-ten­ant or sin­gle-ten­ant archi­tec­ture, each brings its ben­e­fits and neg­a­tives. Con­sid­er cost, effi­cien­cy, hard­ware, and soft­ware access com­par­i­son to help make the well-informed deci­sion.

Additional Resources

If you enjoyed this arti­cle, I rec­om­mend join­ing my email newslet­ter. You’ll be noti­fied when I pub­lish oth­er arti­cles and help­ful guides for improv­ing your SaaS busi­ness. Sub­mit the form below to sign up. Also, use the email icon below to share this arti­cle with some­one else who might find it use­ful.

If you’re the founder and CEO of a SaaS com­pa­ny look­ing for help in devel­op­ing a dis­tri­b­u­tion chan­nel strat­e­gy, please Click Here for more info.

Yes, I want to receive free arti­cles on
How to Scale and Grow a SaaS Busi­ness
 
First Name *
Email *

This form col­lects your email so that we can send you the free mate­ri­als you request­ed. Check out our Pri­va­cy Pol­i­cy for details on how we pro­tect and man­age your sub­mit­ted data.

Facebooktwitterlinkedinmail
author avatar
Vic­tor Cheng
Author of Extreme Rev­enue Growth, Exec­u­tive coach, inde­pen­dent board mem­ber, and investor in SaaS com­pa­nies.
See Full Bio

Related Posts

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top